Monotreme, You seem intent on beating this one to death with regard to the seond sequence at GenBank. I don’t think you are really scoring many points. The latest paper you cited has two Tables. You indicated that you were confused because there was no accession number in Table 1 and the legend in Table 2 was ambiguous, because although the accession number for the new sequence was listed, the old accession number fell within the range listed for most of the sequences AF102657−70.
However, table 1 lists the cloning history,
HK/516/97 60 Female MV; died C2E2 A NXA <101.5 High
and the history for A/HK/516/97 is C2E2, and as I indicated earlier, that is the cloning history for the second sequence, as listed in the record at GenBank and Los Alamos. Thus, there was no amiguity in the isolate used for the mouse data.
For table 2 there was only one position (668) in NA that was listed, and that position is the same in both submissions from patient 156, so technically both accession numbers are represnted on the same line because there is no difference between the two sequences at position 668.
Thus, I don’t think your comments are harsh. I think they are inaccurate and misleading.
Am I missing something?
Similarly, your statement that the authors submitted a new sequence under a new accession number, but didn’t know the new sequence was different also doesn’t make sense. If the authors thought the sequence was the same, there would be no reason to submit a new sequence (the old sequence had already been submitted and published).
Please explain your logic. At this point I really don’t follow.
The patient number menbtioned above in my comments should be 516 (not 156)
Monotreme, I think a review of the data is in order. I suspect that on the publication front, you have reached the three strikes and you are out level. You are fixed on two NA sequences from patient 516 who ddied in Hong Kong in 1997 after being infected with A/HK/516/97. By convention, the H5N1 from that patient is named by location, patient number and date of isolation.
The Nabauro lab at the CDC submitted the original NA sequence from H5N1 from that patient in 1998 and wrote up a paper on the H and N sequence. In 2000 they submitted a second sequence from H5N1 from the same patient. They did not change the number of the isolate, but did submit a new sequence to GenBank, gave the new sequence a completely different accession number, and indicated that the new sequnece came from H5N1 that had been cloned and passaged through MDCK cells twice as well as chicken embryos twice. In association with the new sequence, they also published a paper that characterized the isolates when put into mice. They then published a third paper, on the other 6 genes, noting similarities between the sequences in birds (H9N2 and H6N1) and people (H5N1).
You proposed that the lab didn’t realize that the second sequence was diffeent from the first. The logic behind such a statement escapes me, because if the lab didn’t know the sequences were different, they would not have taken out a new accession number with the new sequence and a passage history.
However, you also complained that the two sequences with the same name would create confusion in the publications using the sequences because the reader would not know which sequence was being discussed.
I indicated that the authors had taken out a new accession number, so the accession number eliminate any ambiguity.
We have now gone through the three publication, and as indicated, the authors took care to eliminate amiguity. The 1999 publication came out when only the first sequence was public. However, that paper used the accession number of the first sequence, as expected. The second paper came out after the second sequence was deposited, but the second paper used the second accession number in the legend to one table, and included the cloning history in the legend to the other table, so again there was no abiguity because unique information (activity in mice) clearly used the second sequence. The third paper did not use either accession number, because it did not discuss the NA sequence. That paper was focused on the other six genes from A/HK/516/97.
Thus, at this time your are 0 for 3 on the amibiguity front. None of the three papers had any ambiguity. The first paper used the first sequence, the second paper used the second sequence, and the third paper didn’t use either.
On the “authors took out a new accession number but didn’t know the sequence was different” front, you position simply lacks logic or an understanding of why sequences are made public and linked to a unique identifier.
Maybe you could clarify.
niman at 17:08 :
>GSGS, There really is no reason to put the second H5N1
>in anyone but the father. The H5N1 from the father had
>21 polymorphisms not in the consensus. Only one was in his son.
>Thus, the fathers H5N1 was mostly a second H5N1 with one
>polymorphism acquired from his son. The father had one H5N1
>from his son, and a second distinct H5N1. The H5N1 isolated
>from the father had the 20 unique changes and one shared with his son.
>In general, your scenarios are far more complicated than required
>to explain the data. I suspect that is how you came up with 200
>((or more than 200 changes). You should review your approach
>toward data analysis when the results you generate have no
>relationship with reality (and you should not assume others
>use your logic).
apparatly you still didn’t understand that argument,
else you would construct a (strong) case for dual infection
+recombination from it. Much stronger than your argument
which only considers the Karo mutations but ignores the
other sequences in Indonesia.
But as I said, only if the (preliminary) mutation data are correct
and we don’t assume hundredths of differences from the
other Indonesian sequences.(which the phylo-tree refutes)
niman at 05:31 :
>…
>However. Monotreme seems to think that two sequences from
>the same patient are a low probability event. He hasnt
>really said why, other than to allude to two independent
>infections, which seems to not only require two infections,
>but two infections by a person. I am sure that few would
>argue that such a scenario is highly unlikely since there
>were only 16 reported human cases in Hong Kong.
>I really know of no viroligist that would take such a narrow view.
>In fact the main point of both cited papers was the fact that
>indeed the 16 human infections were from birds infecting humans.
>There was no itermediate host. This data is supported by the
>sequencing data in the birds and people. Both data sets had a
>number of distinguishing characteristics.
so, the double-infection in a human with 2 different strains
of H5N1 which was “highly unlikely” in Hongkong is OK now in Karo ?
(16 human infections in Hongkong population=7million,
~80 in Indonesia,population=200million)
The double infection in a bird and then to indexcase—rest of cluster
indexcase—son—father has the problem that only the father
had the 2nd strain and 6 others had few mutations.
The case with the 2 Vietnamese sisters with different viruses
sounds interesting…going to search for it…
NS1:
If it truly is simple, it’s by virtue of crisply defining “dual infection” and “different strains”. This can be accomplished by assigning a value to the evolutionary distance between isolates, but some decision must be made as to the optimal way to measure this. Where strong claims rest on the concept: “dual infection”, there is an implicit requirement to immobilize the goalposts by specifying both the criteria and the values used in making the distinction — especially when differences so small as to be easily “mistaken” for random mutation are attributed to recombination.
One of these days, we really might benefit from a discussion on the differences between opinions, informed opinions, reasoned judgements, and personal beliefs. Here, I’d say we’re looking at conventional wisdom: a consensus among informed opinions and reasoned judgements. An opinion on the recombinomics issue can’t be upgraded to an informed opinion unless one is informed on such details as the criteria used in deciding when a “variant” becomes a “strain”.
pandemicflu, The father was infected by his son (who he held closely) and a second H5N1 from a source similar to the source that infected the index case. That is why the father;s H5N1 was so different than the other cluster mebers.
unlikely.
OK, let’s try again.
The paper we are discussing is:
Molecular Correlates of Influenza A H5N1 Virus Pathogenesis in Mice
They list two accessions in table 2, AF102660 and AF296752.
These two sequences are very different from each other. They code for proteins with non-conservative changes. The authors did an experiment with mice with one of the strains represented by these sequences. I agree this is most likely AF296752. However, they gave no indication that this sequence came from a different virus than AF102660 represents anywhere in the paper. If I’m following you correctly, Dr. Niman, you are asserting that the CDC realized that the patient originally described in the Bender et al paper back in 1999 was actually infected with two very different H5N1 strains - at aome point. Any idea when they discovered this? Further, you are implying that they did not find this worth noting in the Katz et al paper in 2002 when the second sequence was published. Further, you seem to be implying that they decided to use the second virus for their animal experiments, but decided not to test the first virus in a similar way. Finally, you seem to be implying that it is perfectly OK to make statements about the pathology in a patient who was co-infected with two different strains of H5N1 based on animal experiments with only one of the viruses. Do I have your interpretation right?
Scaredy Cat at 02:34:
That suspension has been lifted. We had a conversation as I’d hoped we would.
I would assume that the discovered the new sequence after they cloned the H5N1. That is the virus they obtained after cloning and passage. Since it was different, they submitted the new sequence, and that is the H5N1 used in the mouse experiments. They show the passage history in the paper as well as the new sequence submission sheet.
Resequencing of H5N1 after passage is not that unusual, and since the virus that grew was different than the original sequence, that is what they used. Plaque purification of virus is common. Assuming that only one H5N1 is present in an organism is foolish. If you recall, the delay in data from Canada was also due to cloning requirements prior to serotype and sequencing analysis. The passage history of 2 times in MDCK and 2 in chicken eggs is given. I am not sure why the CDC grows human H5N1 in chicken embryos. It is my understanding that they do this with the human isolates in Indonesia also. In the past HKU just used MDCK. I am not sure if they have changed also, but growth in chicken embryos could limit detection of receptor binding domain differences.
However, I suspect that if you e-mail the CDC, that can tell you how long the H5N1 is grown in MDCK cells or chicken embyos, and they can probably also tell you when they did the second NA sequence. I don’t epxect them to say “what second sequence”.
If I had to guess, I would say each passage was for about a week, so my guess is the H5N1 was grown for about a month before they did the mouse experiments.
Table 1 shows the passage history of the H5N1 used in mouse experiments. That history (C2E2) matches the second sequence.
Ask the author? Are we allowed to do that? Wouldn’t that be asking for directions?
Monotreme at 14:05 First author’s contact info from “paper we are discussing” Molecular Correlates of Influenza A H5N1 Virus Pathogenesis in Mice
Mailing address: Influenza Branch, Mailstop G-16, Centers for Disease Control and Prevention, 1600 Clifton Road, Atlanta, GA 30333. Phone: (404) 639–3591. Fax: (404) 639–2334. E-mail: JKatz@cdc.gov.
Sonny, I would say asking the author would be something useful for anyone who would think that cloning an isolate, sequencing it, submitting the sequence to Genbank, and publishing the resukts is somehow anamolous. So far, the only person who has expressed this view has be Monotreme.
My suggestion is that he do that and issue a public apology to the lab for creating misconceptions concerning the sequence they generated.
He should then do the same for the Olsen lab. He is politely accusing them of gross negligence by submitting 56 swine sequences with dozens if not hundreds of lab errors. He indiectly made this accusation over a month ago, and refuses to address the issue, other than a general comment that there can be errors in lab results.
Instead he has spent the last couple of days trying to build a case, which I think he has failed to make on a single sequence that was submitted with a second accession number by a very competent and well respected lab that has published unambiguous results using a cloned H5N1 with that new sequence.
niman at 16:28:
He should then do the same for the Olsen lab.
The only time Monotreme has even mentioned the Olsen lab by name is when he was quoting you. I checked. Find me a quote that contradicts me.
He’s indicated that he thinks there’s a problem that’s worth exploring — and that he intends to invest his own time in exploring it — and you’ve inflated that into a direct public criticism of a lab based on data Monotreme has already stated he hasn’t looked at yet. You’re the one naming names before the investigation’s barely begun and you’re condemning somebody else for doing it. Don’t be silly.
pogge, This really isn’t a new issue. Here is a post from over a month ago, where Monotreme had looked at the sequences and determined that comments on the sequences having large regions of identity to be accurate. Here is one of his posts from over a month ago:
http://www.fluwikie2.com/pmwiki.php?n=Forum.AccuracyOfFluPolymerase
More recently, he has cited lab error (and all of the sequences came from the Olsen lab), so he really is cricizing the lab.
He hasn’t spelled out how 50–100 errors could be made, but he has pretty much zeroed in on that lab.
niman:
All I see in that thread is requests for information, not public charges.
You keep ending up in the position where you’re not just disagreeing, you’re telling someone else why he should sit down and stop talking. That’s why you and I end up talking. Either people don’t know enough or they don’t have the credentials, or what they’re doing is dangerous or now, what they’re doing is slanderous even if you’ve made up the slander and put it in someone else’s mouth.
You’ve pointed to one possible outcome of an investigation — the most scandalous you can think of — and you’re pretending that Monotreme has voiced that conclusion so you can condemn him for it.
OK, well it looks like I’m not going to get response to my post at 14:05. So, we can assume that my statement of Dr. Niman’s viewpoint is correct.
He believes:
Saying that the discovery that a patient was infected with two different strains of H5N1 requires no notice at all by the CDC boggles my mind, frankly.
Allow me to disagree. I would think the discovery that a patient was infected with two strains of H5N1 would be big news. Might even warrant a press conference. But I would think it would warrant at least a mention in a paper where a second virus was used in an animal study to explain the pathogenicity of H5N1.
Actually, Dr. Niman’s version of events is far more damning than mine.
Thanks, pogge.
I appreciate your moderation as I have taken a blood oath not to respond to any calumny by Dr. Niman ;-)
Let’s see how long I can keep it.
Hey you wiseheads…stop arguing, please. We need some productive discussion here, uh?
Thinlina,
Read the earlier threads. There is a lot of history here which predates you.
Monotreme, I don’t think many virologists would be surpised by two H5N1’s in one patient. Based on your comments, I’m not surpised that your mind is boggled by such a possibility in Hong Kong in 1997.
I think it’s time to move on.
Do you have an expalnation for the Canadian swine sequences from the Olsen lab? A month ago you verified that the Candian swine sequences had sequences that match earlier published sequences. I described the data for PB2 and PA on March 18 and March 20.
There are similar examples in the other 6 gene segments, but why don’t we start with the 14 swine sequences for these two genes, which you have had a chance to review?
Some time ago Garden Spider emailed Yi Guan in Hong Kong and posted it at curenevents. We were so thriled. why *we* never followed up, I don’t know.
Do you know of Dark Horse? She has an interest in pandemic flu on her web site. I think she’s “Court TV”. Wouldn’t it really be something If you guys and her could get to get together and do some small telle conference videos between a pannel of 0fluwiki’s and a guest author or a couple of scientist for a QandA.?
Dr. Niman,
Well I take your response at 19:16 to indicate that you agree with my characterization of your position at 18:39. Given that, I agree it’s time to move on, although it would be nice if we could get a comment from the CDC on A/HK/516/97. Non-virologists like me would like to know if the official position of the CDC is that people are being routinely infected with two distinct strains of H5N1.
Lets ask the CDC
Before we get on to the important subject of H1N1 in swine, there are a few more H5N1 cases in humans I’d like to discuss. For example, A/Hong Kong/532/1997. There are two different sequences for the polymerase A gene provided: AF257199 and AJ291402.
There are 21 mismatches between the nucleotide sequences. At the protein level there are 11 mismatches.
Perhaps we could save alot of time if you’d tell us if this is another dual infection, in your opinion.
I think the WHO response to the Karo cluster was a clue. Although the father of the nephew had 21 differences with the consensus of the cluster, they considered such differences as being not important and not worthy of comment in their characterization of the seven H5N1 sequences from the cluster.
Well, the WHO is not the CDC, but I take your point. For myself, I would think that 21 differences is worthy of comment and is quite important, but that’s just me, a non-virlogist.
thanks for the explanations of the dual infections. (it deserves an extra thread ?) I’m still uncertain what to think about the mouse-brain mutations. I’m not so much interested in this 516-patient…
there are also 2 strains A/Hong Kong/486/97. They differ by
4+2+2+0+3+5+0+0=16 nucleotides.
And 2 strains A/Vietnam/1203/2004, they differ by 2+2+0+1+0+1+0+0=6 nucleotides.
A/Ck/HK/31.4/02 differs from A/Ck/HK/31.2/2002 by 144+124+200+37+124+40+38+35=742 nucleotides (?)
here are the strains for the mouse-brain experiments:
131 A/Chicken/HongKong/FY150/01(H5N1)
132 A/Chicken/HongKong/FY150/01-MB(H5N1)
133 A/Pheasant/HongKong/FY155/01-MB(H5N1)
134 A/Pheasant/HongKong/FY155/01(H5N1)
135 A/Chicken/HongKong/NT873.3/01(H5N1)
136 A/Chicken/HongKong/NT873.3/01-MB(H5N1)
137 A/Chicken/HongKong/YU562/01(H5N1)
138 A/Chicken/HongKong/YU822.2/01-MB(H5N1)
139 A/Chicken/HongKong/YU822.2/01(H5N1)
nucleotide-differences, sum in all genes:
---|131 132 133 134 135 136 137 138 139
---------------------------------------
131|--- 020 232 227 232 218 410 511 526
132|020 --- 220 216 242 226 412 503 522
133|232 220 --- 055 227 211 359 421 442
134|227 216 055 --- 217 201 348 420 449
135|232 242 227 217 --- 024 391 481 501
136|218 226 211 201 024 --- 398 485 503
137|410 412 359 348 391 398 --- 280 303
138|511 503 421 420 481 485 280 --- 068
139|526 522 442 449 501 503 303 068 ---
-----------------------------------------
pandemicflu, Thanks for the table. It illustrates that indeed H5N1 in Hong Kong was a mess in 2001 as it was in 1997 (all birds in Hong Kong were culled in 2001 also).
However, the diiferences provided a roadmap on just hon the game is played (and it is NOT via random mutation). The complementary nature of the co-circulating groups showed how H5N1 uses tempate switching do make designer genes. The data is layed out in the filed and http://www.recombinomics.com/patents.html}public PCT (and more will be public within the next 6 months, including the Canadian swine and who the parents are).
Recombination is the name of the game, and H5N1 is playing with a full deck. Those citing “random mutations” need to learn how to read.
The correct link to the road map.
Recombination is the name of the game, and H5N1 is playing with a full deck. Those citing random mutations need to learn how to read.
Is there one virologist, other than yourself, who is “literate”?
Perhaps you are right, and every other virologist is wrong, but people should know that your jibe isn’t just directed against me, but against all virologists that have expressed an opinion on this.
Explaining dual infections of H5N1 in birds are not a problem, dual infections in humans are. The reason is that there are large numbers of birds who have been infected with H5N1 and relatively few humans who have been infected. The odds that an individual will be infected with two different strains of a given virus obviously goes up if that individual is exposed to many other individuals with different strains of H5N1. Thus, areas where many species of migratory birds meet would be an excellent place to acquire multiple strains of H5N1. I have no doubt this occurs and that it leads to reassortment, and on occasion, recombination of H5N1 viruses. This is conventional science. However, for a human to be infected with two H5N1 strains requires more fancy, and in my mind, statistically less likely events. Separate infections of humans with different H5N1 strains is not believable, IMO. The alternative, presented by Dr. Niman and NS1, is that people the birds carrying two strains infect human patients with both strains after one exposure, the same exhalation, as NS1 puts it. I find this equally hard to believe, especially in multiple humans. Although I am sure there are many birds with dual infections as an absolute number, I suspect this is a relatively small percentage of the total populuation. Thus, the odds that a given bird has dual infections is likely small. Further, the bird would have to be producing large numbers of both viruses at the same time. Finally, both viruses would have to be successful in infecting a human after a single exposure.
I have identified two cases where 2 different viruses have been assigned to the same strain. pandemicflu may have identified others. How likely is it that multiple humans have been dually infected with different H5N1 strains from “single exhalations”? Since the submitters of the sequences, including the CDC, have not commented on individual patients being infected with multiple strains of H5N1, it would be interesting to find out if they interpret their data the same way Dr. Niman does.
Monotreme, My comment on reading skills was not directed to anyone in particular.
Monotreme, Earlier I linked the nuerotropism study on H5N1 in Hong Kong in 2001. The data suggested a significant percentage of birds were infected with multiple H5N1’s, and in Hong Kong in 1997, there was close linkage between H5N1 sequences from birds and humans.
I don’t think you will find many virologists that would commit to a single flu entity in birds or people, expecially H5N1 in Hong Kong just prior to massive culls.
As far as reading skills go, most virologists can see the recombination in some types, like picornaviruses or coronaviruses. For flu, most of the achnkowledgement at this stage is in mainland China. However, even they have problems with the “point mutations”.
Next month I will give a reading lessons at the Vaccine Meeting in Cambridge in the morning, and then give a pop quiz in the panel discussion in the afternoon. Since a representative of the Bill and Melinda Gates Fiubndation will also be on the panel, I’ll try to detail the lack of literacy worldwide.
Hi, just want to say thank you to all those who work so hard to make sure this stays on topic. I’ve been reading this with interest, but multiple PC crashes and family obligations mean I haven’t had time to look up the stuff that’s being discussed so I’m not going to make any comments till I’m able to do it properly. Thanks all!
sorry, that was me. This is my daughter’s laptop.
Niman states categorically that it happens all the time and that it’s meaningless to credentialed virologists. Only interesting to “outsiders.” I strongly doubt this. I don’t think there can be any argument that this occurence would be significant and of critical interest to the scientific community.
Here’s the possibilities so far:
1. Niman’s explanation as above. Multiple H5N1 strains isolated at different times from one single sample from one patient. Coinfection, happens all the time, no big deal. As a proof of Recombinomic Theory, one would think that Niman…okay, well, everybodywould be all over that! Boy, if I was Niman, I’d be writing that up so fast for publication, it’d make your head spin! As a matter of fact, it would’ve been done this weekend!
2. Administrative error, mismatching sequences and identifiers. In the lab, data entry clerk, whatever. Not noticed because people familiar with that study and that sample wouldn’t have expected to see multiple different sequences, so they didn’t do any sequence aligments, thereby seeing the differences.
3. They were not aware of different isolates from the same sample.
4. A change in the sample happened over time, for whatever reason. The original sequence may not be present in the sample at this time and may not be able to be isolated. Multiple passages, whatever.
glo at 14:45:
one would think that Nimanokay, well, everybodywould be all over that! Boy, if I was Niman, Id be writing that up so fast for publication, itd make your head spin! As a matter of fact, it wouldve been done this weekend!
I’d like to address more than just this comment. With no disrepect intended to anyone for the, um, earthy way I’m putting this: conversations about niman and speculation on his motives is a dead horse that’s been beaten so long the poor, bloated carcass should really be sent to its final resting place before it becomes overly ripe. These discussions go off the rails when they get personalized like that and when you start talking about someone who’s here in the third person, you should probably think twice before you post.
You seem to be expressing the same surprise as Monotreme at the fact that a dual H5N1 infection in a human patient nine years ago was such a unnoteworthy event that you’re only finding out about it now in a casual discussion on a message board. You seem to be suggesting that such an event would have been publicized by niman himself and you’re wondering why not. Just ask.
Does that restate with personalities removed?
Dr. Niman:
Since Monotreme is trying to focus on human sequences, and you’re going to present at a vaccine meeting soon, I would like to ask a question. You mentioned the dual infection in the Vietnamese groom who infected his two sisters with a different strain, one GI, one respiratory. Yet, the CFR remained the same. Wouldn’t a dual infection be more virulent than a single infection? And, if it is even an average amount of time that an exposure results in a dual infection would this effect a single vaccine development, or require two seperate versions of vaccine?
It appears to me that perhaps both Monotreme and Niman are correct.
Given that there is apparent consensus that there are indeed different strains of H5N1 circulating in birds, and that there is typically, at least in the index case in most clusters, prolonged intimate contact with birds, what is the likelihood of simultaneous infection by 2 strains of h5N1? I wonder why this seems so unlikely? I would assume, given the situation in Indonesia, and indeed in China, that H5N1 is more or less endemic in the bird populations there.
If the virus is a sloppy replicator, then it would seem likely that a single infected cell would produce many different varieties of virus. What are the likelihoods that 2 separate strains of H5N1 could produce mutations requisite for infecting humans? Probably pretty low.
Once in the human, the likelihood of recombination would seem to be higher, no?
Not saying that’s the way it happens, but I would be interested in your thoughts.
Monotreme at 09:23
Let’s be certain that all readers know that I’ve mentioned a potential event (dual infection of a human from single exhalation from a lower-order Influenza host organism) for purposes of illustration, to border the conversation.
This single exhalation, descriptive and illustrative event is not the most likely method of multiple infection. Many other come to mind that are more likely.
In a world where humans live in close contact with animals and the animals do not observe proper usage of fecal removal facilities (toilets), the humans (handlers and those in casual proximity) will most certainly be regularly walking through vestiges of feces and aerosolised feces from the pressurised spraying of water to clean the animals areas. A portion of those humans will inhale infectious particles. Some group of those humans will fail to properly sanitise their shoes and clothes prior to returning to their homes. Some of those humans will not wear shoes at all and will have slightly abraded feet. Some of those humans who inhaled particles and failed to sanitise their shoes will use their hands to remove their shoes and then may sneeze from the dust particles that are stirred. If a tissue is available, they may grap a tissue with their now contaminated shoe-removal hand and wipe their face or eyes and then flip the tissue and wipe again, spreading the contaminant. If they don’t sneeze, they may go to the sink to wash their hands after shoe removal and leave feces particles on the soap dish for later consumption.
You can see that proximity = opportunity, an age-old epidemiology maxim that continues to hold true here.
Put on your thinking caps and you’ll each see that if a flock of birds have dual infections of H5N1 (replicator extraordinaire) and is near people that the odds of dual infection transmission are not much higher than single transmission.
Same vector, same host, same physical area.
Virus particles are small and H5N1 does not require 100,000 particles to infect. Even if 100,000 were required for each strain, this little . just about covers the real estate required to store that 200,000 count.
So let’s not work so hard to discount a bordering and educational illustration in the future.
Gather and solve.
LMWatBullRun at 16:58
Solid introspection on all counts.
Absolutely, pogge. It’s your website. No offense intended; I was clearly mistaken in the impression from the tone of this thread - the name-calling, insulting intelligence and direct comments about not knowing how to read - that rough teasing was convention and encouraged. I have six brothers and when told that I’m stupid and illiterate, it’s a gesture of affection and I would be remiss in not returning the gesture. My mistake. Cultural difference.
Monotreme, there’s more interesting bits and pieces to this, but inappropriate to a casual discussion on a message board. FYI, I have asked for clarification from a credentialed working scientist. :)
LMWatBullRun at 16:58 If the virus is an incredibly sloppy replicator, then why did six individuals who were serially infected have identical copies of the virus? And why, despite sloppy replication, can we construct phylogenetic trees that show that certain polymorphisms crop up over and over? And why don’t the local birds have RESRRKKR? ++pulls hair++
wetDirt-
You’re right. We cannot fully explain the evidence using the conventional framework?
I was clearly mistaken in the impression from the tone of this thread
This thread has a fairly broad range in tone so even if we call it a mistake I’m not saying it wasn’t an honest one. I’m trying to narrow the range to the right point so that I can stay out and just let people talk. It’s a work in process.
wetdirt, Why don’t ANY H5N1 sequences at GenBank have RESRRKKR, other than the ONE human sequence from Indonesia (and the otehr Indonesian human sequences that are being withheld).
I would say that the thinking about H5N1 evolution and replication is much sloppier than the polymerase.
NS1, sorry if my interpretation of “a single exhalation” was excessively literal. I warned you before, I only understand literal prose :-)
Both you and LMWatBullRun state that exposures have been prolonged. But what is the evidence for that? One of the surprising things about H5N1 infections in humans is who isn’t infected: poultry cullers, for example. I have seen pictures of people sleeping in poorly ventilated rooms filled with floor-to-ceiling cages of chickens in Southeast Asia. Yet, these are not the people who get infected. Instead, we have clusters like the first one in Indonesia where an upper-class family with little or no exposure poultry was decimeated by H5N1. In the Karo cluster, they can’t even find the source of the infection. Doesn’t sound like prolonged contact to me.
I think your phrase “a single exhalation” is closer to the truth than you intended it. There is much discussion regarding genetic susceptibility, which I initially rejected, but now think may have a role in explaining why some people, and their families, become infected. Another factor, may be a mammalian vector that has not yet been identified. In any case, there is no evidence that most of the people who became infected with H5N1 did so as a result of prolonged exposure to animals. The H2H may be a different story.
wetDirt at 17:25
If the virus is an incredibly sloppy replicator, then why did six individuals who were serially infected have identical copies of the virus? And why, despite sloppy replication, can we construct phylogenetic trees that show that certain polymorphisms crop up over and over?
In one word, the answer is Selection.
Mutation rate should not be confused with evolution rate. These are distinct terms with quite different meanings. I think the conflation of these two terms has resulted in a misunderstanding of the role random mutation plays in non-random evolution. Viruses aren’t the only organisms that make mistakes, human embryos do too. Over half of all human pregnancies result in spontaneous abortions. Why? Because of genetic errors. Mother nature is careless with humans and even more careless with viruses, especially influenza. We don’t see many of the mutant flu vires because they are aborted, stillborn or fail to thrive. We only see successful viruses. Most mutations are punished with death. This is called Negative Selection. Some mutations have no effect on propagation. These are called neutral mutations. Every once in a while, a mutation occurs in a virus which allows it to replicate or be transmitted more efficiently. This is called Positive Selection. Thus, the degrees of freedom for evolution are severely limited. This is why you see non-random patterns in the observed polymorphisms. Most of the mutants that have occurred were buried before the virus that harbored them got past the first cell they were born in.
Mother Nature’s motto: “Go ahead and mutate, I’ll make more.”
glo at 17:18
Monotreme, theres more interesting bits and pieces to this, but inappropriate to a casual discussion on a message board. FYI, I have asked for clarification from a credentialed working scientist. :)
I look forward to reading what you hear from your expert. I’m going to be doing more research on the human H5N1 isolates and will be looking for a pattern for the “dual-infections”. Perhaps the pattern will convince me that dual-infections are common in one province in Vietnam, or in duck-blood drinkers. Or I may find that all the “dual-infections” involve sequencing from one particular institution. This is going to take me awhile, but I’ll report back here, when I’m finished.
Monotreme at 21:20,
This is called Positive Selection. Thus, the degrees of freedom for evolution are severely limited. This is why you see non-random patterns in the observed polymorphisms.
Thanks for your clear explanation. I am wondering, is it possible that same or similar selections could occur at considerable geographic distance, or even time distance…possibly an explanation for same or similar mutations that occur a few years apart? Kind of like two inventors living far apart, creating the same invention within a short time span.
I realize this is speculative, but don’t see why it couldn’t happen.
Monotreme, here I disagree with you on the nature of spontaneous abortion. The vast majority of these cases are not due to single-nucleotide mutations, and using this example unnecessariliy conflates a lot of germ cell errors that have nothing to do with mutations. Certainly, single-nucleotide errors account for far less than 50% of pregnancy loss. Loss categories include maternal factors, nutritional factors, immune dysfunction/rejection, and inherited defects. Even Down’s syndrome is not clearly a mutation, it has heritable aspects and links to defective sperm, among others, which are also not point mutations. To get an idea of the human point mutation rate, you have to look at things like all those weird metabolic and immune dysfunction diseases that don’t run in families. And you will find them very rare. Actually, if you look at the pregnancy loss rate for IVF, you would find it very low—and they can’t prenatally screen for point mutations.
We still haven’t resolved the high mutation rate idea. I would be interested in looking at other references you consider ‘good’ that demonstrate (using sequences)that there is, in fact, a high mutation rate. I’ve been around too long to take this on faith. Next, you will have to tell me whether or not the mutation rate is constant under all circumstances, and that there are no confounding factors such as selection caused by growing virus on a substrate that doesn’t match its original one. Like, if it was found in a duck, then grow it in duck eggs, not chicken eggs. I have a funny feeling that the inherent mutation rate is variable, and increases when the virus is stressed, such as when it is being stared at by grad students, or trying to reproduce in a foreign host. When it is in a host that tolerates it better, it may be less stressed, and do a better job. I also suspect the virus does a better job when there are dual/multiple infections, because it can weed out defective bits during assembly. The selective advantage of being able to swap in whole chunks of code to circumvent host defences, or flip from low to high path, is huge—the virus isn’t reinventing the wheel. This also explains why the shift from low to high path is often sudden, not gradual. I suspect that the difference between low and high path is at least a dozen or more nucleotides. What’s the chance that happens by chance? note to self: why aren’t we seeing medium path? what does that mean? is it a neuramindase kinking problem, two metastable states of folding that can flip back and forth, but won’t stay halfway?…++wanders off mumbling incoherently++
“Loss categories include maternal factors, nutritional factors, immune dysfunction/rejection, and inherited defects.”
Actually, the vast majority of abortions are due to fetal/germ line genetic defects, whether hereditary or not.
On re-reading the posts, I think Monotreme’s main point of using abortion is not about whether they are due to point mutations, but that Mother Nature is careless with reproducing any organism, and often they don’t survive.
“”Actually, the vast majority of abortions are due to fetal/germ line genetic defects, whether hereditary or not. “
Well, there’s a ways to go from a nucleotide to a gene…Human genes have a lot of junk in them, but the rate of junk accumulation is different from the rate of fetal loss. Apples and oranges.
Monotreme’s previous post began “Mutation rate should not be confused with evolution rate”.. He is correct in making a distinction here, but then chooses an example that confuses gene-level selection with single nucleotide selection. If we don’t get a little more precise around here in what we are talking about, we will just go around in circles.
beehiver at 21:50
… is it possible that same or similar selections could occur at considerable geographic distance, or even time distancepossibly an explanation for same or similar mutations that occur a few years apart? Kind of like two inventors living far apart, creating the same invention within a short time span.
Absolutely. All possible mutations will happen again and again. Whether they are observed or not depends selection. If the selective force is the same in different regions, ie, the same vaccine is being used on two different continents, mutations that favor evasion of immune response will be selected for independently on both continents. It’s quite possible that the same mutation will be independently selected for.
wetDirt, anon_22 correctly points out that I was using spontaneous abortions, still births and failure to thrive as analogies for the sorts of mutations that can occur in viruses.
I disagree with some of your characterisations of the causes of human abortions, but that’s another subject. The point is: genetic mistakes, whether they are chromosomal or point mutations, occur in all organisms, including humans. We do not see all of the mistakes because they are selected against.
One of the important differences between virus and human genomes is that human genomes contain large stretches of “junk” DNA that serves no purpose. Mutations that accumulate in these regions has no effect on fitness. Viruses, especially flu viruses, have very few regions in their genomes where mutations can occur without an effect on fitness. This is because when coding regions are expressed as a percentage of the total genome, this number is about 3% in humans, and probably close to 99% in flu viruses. Thus, there are fewer degrees of evolutionary freedom available to a flu virus than to humans.
Below is a nice paper that discusses the difference between mutation rate and evolutionary rate with respect the influenza. Note that influenza genes do not evolve at the same rate due to selective pressures from the host immune system.
Codon bias and frequency-dependent selection on the hemagglutinin epitopes of influenza A virus
Although the surface proteins of human influenza A virus evolve rapidly and continually produce antigenic variants, the internal viral genes acquire mutations very gradually. In this paper, we analyze the sequence evolution of three influenza A genes over the past two decades. We study codon usage as a discriminating signature of gene- and even residue-specific diversifying and purifying selection. Nonrandom codon choice can increase or decrease the effective local substitution rate. We demonstrate that the codons of hemagglutinin, particularly those in the antibody-combining regions, are significantly biased toward substitutional point mutations relative to the codons of other influenza virus genes. We discuss the evolutionary interpretation and implications of these biases for hemagglutinin’s antigenic evolution. We also introduce information-theoretic methods that use sequence data to detect regions of recent positive selection and potential protein conformational changes.
in that paper about codon bias, I can’t believe that this bias just only evolves by normal selection and just only by the fact itself that these antigenic codons have and maintain larger volatility. That is too slow and uncertain for antigenic evolution. I feel, that the locations of the antigenic sites are somehow encoded in the genom or that the virus-mutations recognize the antigenic regions from the 3d-structure. Some specific mechanism how the virus realizes that it should mutate differently in the antigenic epitopes.
as for the double-infections , I’d like to see the sequences of the two
Vietnamese sisters with different strains and the brother .
Ngo Le Hung (31M)
Ngo Le Hanh (30F)
Ngo Le Hong (23F)
from Thai Binh, January 2004. How can I find them ?
when there are dual-infections in humans with two H5N1 transmitted from the same host, shouldn’t we also expect dual-infections of two H3N2 transmitted simultaneously, of H1N1 and H3N2, of Rhinovirus and H3N2 or two different common_cold viruses transmitted simultaneously with the same droplet etc. ? Dual infections still should be rare. One bird has virus A and virus B, then they mix and we have 3 vaiants: A,B,mixed (various ratios) and these variants compete. The mix can only survive, if both viruses succeed to replicate but then selection would prefer one strain probably. I see no reason, why selection should prefer the mix ? So the mix won’t increase the product of the two infection probablities per disease-duration. Unless there is some cooperation between the viruses which makes the mix more successful
pandemicflu, I am working on this page to correlate sequence data with specific patients to address the questions you ask. I am interested in the answers as well. However, this work is extremely time-consuming and my time is limited. Ideally, there should be a searchable database to address these questions, but there is no indication that such a database exists.
In my comment above, I was referring to this page:
pandemicflu at 04:11
The mix can only survive, if both viruses succeed to replicate but then selection would prefer one strain probably. I see no reason, why selection should prefer the mix ?
I agree with this. The host is the environment in which viruses are evolving in real-time. The same logic also applies to mutants. The newly formed mutant viruses must compete with normal viruses, perhaps within the same cell, but certainly in close proximity. Most of the time, the mutant virus fails in this competition, which is why we don’t observe them, “in the wild”. When samples are taken from patients or animals, we are primarily looking at the “winners”.
Monontreme,
Yes, the sequence database has the winners. I gave you the winners from Hong Kong in 2001 and 2002, which included multiple examples of different H5N1’s from the same source (some defined by passge through mice, some defined by cloning, and some implied because of multple rounds of cloning (several are dot 4 or dot 9).
The same thing is true for the Canadian swine. Over a month ago I have you the clear examples of 1977 genes from two different H1N1’s in PA and PB2 genes along with well characterized isolates from 1998 and 2002. These all had extensive regions of identity, integrated at multiple sites in the 2003/2004 swine genes.
You agreed that the descriptions and identities were accurate and then suggested that these results were some sort of lab error(s). However, specifics of errors have have been forthcoming on the two genes described in March. The descriptions define the parental stains as well as the precise locations where they had intergrated into the Canadian swine genes.
You still haven’t addressed how these “winners” were created. I have estimated that somewhere between 50–100 lab errors would be required to explain PB2, PA, and the other 6 gene segments repressnted by the 56 sequences at GenBank.
pandemicflu at 02:03-
I am curious why you think codon bias cannot be a simple artifact of selection. Please elaborate your thinking on this subject, if you would.
Monotreme-
If one accepts that the vast majority of the Indonesian population are in constant contact with infected birds and that they have not been infected (which case I make solely for the sake of the argument; I would be *most curious* to see the results of a thorough and competent seroprevalence study), then if a group of people had a genetic predisposition to infection by one h5 virus, why not another? Would the predisposition to infection by H5 be that sequence-specific?
Wetdirt- Further, once the dual infection and recombination occured in the index case for the Karo cluster, I *would* expect to see only one successful viral variant thereafter, as Monotreme put it, ‘the evolutionary winner’ infecting that particular human genome. I think it would be illuminating to compare the genetic structure of those infected in Karo with the structure of the attending HCWs and see if we can identify the predisposition. And yes, I DO realise how large a task that is, even with computers, but I’d think that would be a very significant bit of information.
niman, I haven’t examined/don’t remember the 1977 genes, but…
couldn’t it be, that the Tennessee/77 was conserved in St.Jude’s lab,
maybe even recombined there in an experiment and then was accidently
infected to a swine in 2002, while they took another sample
from that farm ?
LM..,we usually only have a few mutations from chicken to chicken
or human to human. We have some positions where apparantly no mutations
occur at all, not just 1.4 times fewer as suggested by the paper.
All codons occur in all regions, not just that some codons
are reserved to the antigenic sites.
One mutation in the antigeneic site can easily change the codon-volatility,
volatility is poorly preserved.
LMWatBullRun at 10:11 “Monotreme-
If one accepts that the vast majority of the Indonesian population are in constant contact with infected birds and that they have not been infected (which case I make solely for the sake of the argument; I would be *most curious* to see the results of a thorough and competent seroprevalence study), then if a group of people had a genetic predisposition to infection by one h5 virus, why not another? Would the predisposition to infection by H5 be that sequence-specific?”
Interesting that you mention host susceptibility. I don’t know the answer to your question but FYI host factors in the Indonesian cluster was also discussed here at 11:50
pandemicflu, The identities were far more extenisive than a single isolate or single gene. I used PA and PB2, because they were easy to write up and see the difference. The TWO isolates from 1997 were discussed the most because of the long time differential, but there were more isolates involved just with these two recent genes. Five of the seven PB2 genes had identity with swine/26/1977. Six of the seven PA genes had many regions of identity with swine/24/1977. The two swine isoaltes were similar to each otehr, but were distinct, and those distinctions were maintained in the 2003 and 2004 isolates. However, regions not having identity with one or the other 1977 isolates had identity with a 1998 isolate from North Carolina or a 2002 isoalte from Korea. Moreover, one of the secent swine isolates had significant regions of identiry with the 5′ half of a 1931 isolate.
Thus, there would have had to have been at least 5 different escapees (1931, 1998, 2002, and two from 1977). They would also have had to integrate with the swine at different positions and for varying legnths. Moreover, the swine were from farms in Ontario and Alberta, so the escaped H1N1’s would have had to distribute over a wide geographic range and continue to spread for over a year since some isolates were from 2003 and others for 2004.
The above discriptiopn is for 2 of the 8 genes. There were similar regions of identity in the other 5 swine genes (PB1 was like human PB1 from the mid 90′s), which have regions of identy with other isoaltes from the 1990′s, so there would have to be over a dozen escapees which recombined with different genes in differnt locations with various anumals spread over at least provinces and over at least 2 years.
“anonymous” made this same suggestion a couple of days ago, and received the same answer. The same questions again and again gets annoying, even if the same question comes from different handles or from different boards.
anonymous posted as GS or GSGS on various boards, and all of thoses posts sounded much like yours. It might be useful to read those questions and the answers I gave rather than ask the same questions yet again.
anonymous made this same suggestion a couple of days ago, and received the same answer. The same questions again and again gets annoying, even if the same question comes from different handles or from different boards.
As a point of information, no one is posting in this thread any longer as “anonymous” or under multiple handles. I’ve asked for cooperation and received it (and I’m grateful for it).
Obviously there’s nothing we can do about what might happen on other boards.
you needn’t answer when you’re annoyed. I didn’t remember an answer like the above one. Others might find it useful ?! and can also elaborate… Even if asked twice, isn’t it worth it ?
I don’t understand the two escapes from 1977. One is enough and in 1988 ? Tennessee/24 and /26 are similar, for 25 we have all 8 genes, St.Jude maybe has all 8 genes for 24,26 but don’t show them. The 1931 and Korea was another event ? May the recombination have occurred in a lab. ? (everyone may answer/speculate…)
here niman’s answers from earlier in this thread again:
The paper on the PA gene is here, so the gescapeeh would have to recombine with the isoaltes below and in different places. However, the isoalte for PB2 was a different H1N1 from 1977, so that one would have had to escape also. However, most of a 1998 isolate from North Carolina was also a exact match with one of the Candian sequences, as was a 2002 isolate from Korea with another Canadian sequence, so the data would require four gescapeesh just for the PA and PB2. In addition, one of the Canadian isolates has extensive homology with the 5 end of the 1931 isolate listed below, so that one had to gescapeh but change somewhat and tehn recombine. For the other genes, more gesapeesh would be required and they all had to get into many different swine on many different farms in southern Canada.
The bottom line is their paper has OBVIOUS examples of recombination, which they maintain was not an artifact, but still failed to mention the recombination in their peer reviewed paper (and probably still maintain that there is no recombination - because it was in all samples and virtually all 8 gene segments and failing to mention it would be a gross oversight).
they could have been experimenting with the strain, recombination in a lab. they might have infected swines deliberately and then it went out of control. Why else should they avoid to comment about the recombination and conservation ? That’s unusual. Recombination doesn’t well explain the conservation either.
Of course, I’m speculating…I’m not saying how likely I consider this.. ah, well. OK. 30%.
Monotreme at 00:00
I just love the bit about “certain genes, and specific residues within those genes, experience frequency-dependent selection to change, whereas other genes experience purifying selection to remain fixed. “ Sorta like one foot on the gas, the other on the brake.
Right here I see a problem. We have to invoke divine providence here to get rid of unwanted point mutations. OK, we will do it by something called ‘purifying selection’(PS). It kind of fits with what we have seen that certain long stretches of nucleotide are highly conserved, particularly in H1N1, but it makes me uncomfortable: Why don’t we see more variety in the silent spots? You’d expect, if PS was involved, the phenotype stretches would have piles of silent mutations, but they don’t have that many. This argues against a high mutation rate in the NA parts. In any case, this is a highly testable hypothesis, all we have to do is look at the diversity of the quiet bits. I don’t think he proves his case here very well, in fact, this looks like deus ex machina to me.
So this article doesn’t do it for me, either. But it is a nice analysis anyway.
Monotreme at 08:42 pandemicflu at 04:11
No reason exists to conclude that one viral strain will win via natural selection or any other theorised framework even within a single organ.
Going farther with multi-tropic H5N1, consider the case of a simple spleen and lung infection that will very likely host variants.
These notions of a winner are somewhat overwrought and have little application to this discussion.
pandemicflu, You think that there is a 30% likelihood that the lab managed to cause recombination in the 16 gene sequences submitted and reported those lab recombinants as examples of infections in swine, or it is 30% possible that they infected the swine and then when they went to collect samples because the pigs will ill, but forgot that they had experimenatally infected them?
Earlier, you (as anonymous) said you contacted the Olsen lab. Unfortunately, the name recombinomics appears to have come up. Since I have yet to see a correct summary of what you think I posted, having you communicate with anyone regarding what you think I said is unfortunate.
So is there something that they said to you that would give your reason to speculate on a public board that they were grossly incompetent and are publishing artifacts of some sort or combinations of a series of errors? Do you suppose the same errors generated the same type of data in the other 5 genes?
As I said earlier, I think it would take something like 50–100 errors if there was no recombination or no conservation of sequence. Are you saying the number of errors is only half of my estimate because the recombination was real, but the sequences were multipel escapees (that were then used to infect pigs on farms)?
Your specualtion requires a very large number of errors, which you are now saying has a 30% liklihood?
wetDirt
D.ex machina. - are you seeing a rules-basis now?
Does so-called selection occur post-facto or pre-facto, pre-emergent?
Post-event limiter or Polymorphism Driver?
Monotreme-
I think you’re still focusing only on my few explicit examples for dual infections and aren’t visualising the framework and the full breadth of these perfect sera and substrates.
I’m Staring at the Pixels and missing the Picasso.
LMWatBullRun at 10:11
“… I think it would be illuminating to compare the genetic structure of those infected in Karo with the structure of the attending HCWs and see if we can identify the predisposition… “
Somebody in the last week or so gave a pretty definitive answer on why the second parental strain couldn’t have been H3N2 from HCW. I hereby drop the idea.
NS1 at 17:32
“D.ex machina. - are you seeing a rules-basis now? “ I’m seeing a lot of ex cathedra arguments in the literature that are a lot like Goodgulf the Wizard’s explanation of how he escaped the Ballhog: Well, once out of the pit…
“Does so-called selection occur post-facto or pre-facto, pre-emergent? “ I’m sure selection is occurring, but I think there is a lot of ‘throwing spaghetti at the wall and seeing what sticks’ going on: Make as many variants as you can with whatever code you have on hand, and fling them out there, and whatever can slip beneath the host defenses wins. I guess you would call this post-facto selection, but pre-emergent generation of different ‘keys’, mainly in the HA genome.
“Post-event limiter or Polymorphism Driver? “
As Gimli said in ROTK, ‘last one standing wins’.
wetdirt- Perhaps I wasn’t clear, but what I meant was the human genetic structure, not the viral. Perhaps the first thing to be done would be a seroprevalence study to ascertain the presence or lack thereof of H5N1 prior infection.
IFF there was no H5N1 in the seroprevalence study then it would be worthwhile to do human genetic studies…..
LMWatBullRun at 19:41 Remember that warehouse at the end of Indiana Jones? That’s where the box with the results of the seroprevalance studies is.
As far as sequencing a random HCW to spot differences from the Ginting family, you will end up knee-deep in differences that are mostly irrelevant, like that one has a gene for curly hair and not the other. The trick if you are going to do this is to find somebody, peferably alive, who is a close blood relative but didn’t get it. Fewer differences, and it is more likely that the differences will be significant. Then you will only have to comb through a million or so pieces of data to find the one you want. Good luck, I’m not holding my breath.
wetDirt
Between 3 and 4 million pieces if you fully subscribe to the intron argument/exclusion. 3 billion or so otherwise.
Small deck, huh?
Wetdirt-
Yup. “Top. Men.” The difference is that we can recreate the data. If we find the right lever to move WHO.
As I said, I don’t think this will be a trivial exercise. However, I do think this would be MOST instructive….Anybody got a couple of grad students handy? <grin>
LMWatBullRun at 10:11
If one accepts that the vast majority of the Indonesian population are in constant contact with infected birds and that they have not been infected (which case I make solely for the sake of the argument; I would be *most curious* to see the results of a thorough and competent seroprevalence study), then if a group of people had a genetic predisposition to infection by one h5 virus, why not another? Would the predisposition to infection by H5 be that sequence-specific?
I agree about the seroprevalence study, but the ones that have been done, and there have been quite a few, have shown no evidence of widespread, asymptomatic cases, in humans.
I think you misunderstood my point about genetics. I don’t think people are genetically predisposed to be infected with one strain and not another. My point was that prolonged exposure is not the key variable. It is likely that very limited exposure to an animal caused disease in susceptible individuals due to the socioeconomic group of some of the victims and the failure to find an animal vector in many cases, including Karo.
NS1 at 15:07
No reason exists to conclude that one viral strain will win via natural selection or any other theorised framework even within a single organ.
Well, I can’t agree with this. A mutant with a stop codon in the middle of the HA gene isn’t going to be very successful infecting the next cell. That’s a loser mutation, in my book.
One strain doesn’t win, all the time and forever. Evolution is dynamic. New strains are born and compete with existing strains. Change the environmental conditions and different strains are favored. Today’s winner may be tomorrow’s loser. Or today’s loser in birds may do very will in pigs, etc.
These notions of a winner are somewhat overwrought and have little application to this discussion.
Well, I can’t agree with this either. Natural selection and luck determines who the winners are. This explains why non-random patterns of polymorphisms are observed. I thought this was what we were discussing? My point is that Recombinomics isn’t the only way to explain non-random polymorphisms.
That being said, neither I, nor any other conventional scientist that I am aware of, denies that recombination occurs. So, providing examples of recombination do *not* disprove conventional science.
What we are arguing about is rank order of frequency.
NS1 at 17:32
Does so-called selection occur post-facto or pre-facto, pre-emergent?
So-called selection? Surely you are questioning natural selection, right? In any case, natural selection obviously acts after mutations occur.
NS1 at 17:43
Monotreme-
I think youre still focusing only on my few explicit examples for dual infections and arent visualising the framework and the full breadth of these perfect sera and substrates.
Im Staring at the Pixels and missing the Picasso.
Darwin, not Picasso. Origin of Species explains things pretty well ;-)
LMWatBullRun at 10:11
“Wetdirt- Further, once the dual infection and recombination occured in the index case for the Karo cluster, I *would* expect to see only one successful viral variant thereafter, as Monotreme put it, the evolutionary winner infecting that particular human genome. “
Here is a place where we differ. First, I am persuaded that H5N1 is heterogenous and in the average duck consists of a few closely related ‘substrains’, as distinct from ‘Strains’, as in the Qinghai Strain (capital S). These substrains differ by a ‘bunch’ of nucleotides. I think it is possible that two parent substrains, one of which contained the key, produced a variety of child substrains. One or more child could inherit the key, and the others would dead-end. As long as the key sequence is there, it doesn’t matter that the other polymorphism are there. It may be that the key was so fiddly at Karo that subsequent generations lost it and deadended. I don’t see any reason for variations to quit arising just because someone won the game.
Monotreme at 23:09
My discussion centers on the viable strains that you consider to be competing for a winner in all cases. Obviously, non-viable strains disintegrate before our discussion begins.
Monotreme-
Have you ever caused a genetic acquisition in the lab?
What is the pre-emergent set of rules that cause a particular acquisition? Are you arguing that all genetic acquisition is random?
Monotreme at 23:15-
If C.D. couldn’t keep the finches straight that he had right before his eyes, how can we demonstrate his view of recombination from that discourse on carrier pigeons? Powerful microscopes in 1861?
NS1 at 03:29
My discussion centers on the viable strains that you consider to be competing for a winner in all cases. Obviously, non-viable strains disintegrate before our discussion begins.
My point is that most mutations will result in non-viable strains. This explains why we observe a non-random distribution of polymorphisms.
NS1 at 03:32
Are you arguing that all genetic acquisition is random?
No. As I have said many times, I believe homologous recombination does occur. This is partially influenced by the amount of homology between the two nuclotide segments involved in the exchange and would therefore not be random. Further, when you say genetic acquisition, are describing only observed strains? If so, my comment above applies. Mutations are random (mostly, there are some exceptions), but evolution is not random. The rules you seek are found in Natural Selection. When H5N1 makes its final adaptation to humans, no microscope will be required to observe the effects of Natural Selection.
I’ve been looking at the genes of the Canadian swine again,
and an escapee or lab errors looks less likely to me now.
Instead I believe, that the mutation rate of some gene-parts
can vary and probably is encoded somewhere somehow in the genome.
Just look at the PA of
A/swine/Tennessee/25/1977(H1N1)
A/swine/Ontario/48235/2004/(H1N2)
A/swine/Ontario/55383/2004(H1N2)
When you compare any two of these three, you will find that there are
significantly more differences in the first quarter of PA than
in the middle half of PA. So the low mutation rate in the middle half
of PA did continue after the escapee infected the swine.
And the polymerase could be very exact, if it only wants and gets
the order to be exact.
It could be that point-mutations don’t occur in the middle
half, or that there are no break-point for recombinations in the
middle half.
That would mean, that any of these 3 viruses above would continue
to show few mutations in the middle part of PA.
I’ll bet that the mechanism isn’t ‘purifying selection’, too. I bet that the third base position is used for error checking.
but then the 3rd-base bits should be evenly distributed, they aren’t. Also in my example above we clearly see regions, a simple error-checking would be region-independent.
Well, let’s assume for a moment that variablilty in PA is directly proportional to the inherent error rate of the polymerase. So the average ‘genetic drift’ rate is to be found in the least mutable part of the genome. There is low selection pressure on this part of the genome, so there are only a few parental strains in the gene pool. In HA, however, there is constant, relentless selection pressure to evade host defenses. Here, a low mutation rate is a huge disadvantage, and it would be better to have multiple parental variations floating around in the gene pool. For HA, having recombination be the dominant mechanism would work well, because the overall gene pool could store the different, well vetted bits of code, and mix and match in the organism itself.
One thing this also rescues is the idea that influenza as a recognized disease doesn’t just vanish. I have been bothered for awhile by the idea that really high inherent mutation rate suggests that influenza can’t retain its own identity for more than a coupla hundred years, because it would mutate itself into chaos.
How on earth can you have ‘purifying selection’ on a gene with low intrinsic selection pressure? It makes no sense at all.
Hot off the press… Look at the last few lines.
http://www.nytimes.com/2006/07/25/sc…syahoo&emc=rss Published: July 25, 2006
Researchers believe they have found a second code in DNA in addition to the genetic code.
The genetic code specifies all the proteins that a cell makes. The second code, superimposed on the first, sets the placement of the nucleosomes, miniature protein spools around which the DNA is looped. The spools both protect and control access to the DNA itself.
…
The new code is described in the current issue of Nature by Eran Segal of the Weizmann Institute in Israel and Jonathan Widom of Northwestern University in Illinois and their colleagues.
… Biologists have suspected for years that some positions on the DNA, notably those where it bends most easily, might be more favorable for nucleosomes than others, but no overall pattern was apparent. Drs. Segal and Widom analyzed the sequence at some 200 sites in the yeast genome where nucleosomes are known to bind, and discovered that there is indeed a hidden pattern.
Knowing the pattern, they were able to predict the placement of about 50 percent of the nucleosomes in other organisms.
The pattern is a combination of sequences that makes it easier for the DNA to bend itself and wrap tightly around a nucleosome. But the pattern requires only some of the sequences to be present in each nucleosome binding site, so it is not obvious. The looseness of its requirements is presumably the reason it does not conflict with the genetic code, which also has a little bit of redundancy or wiggle room built into it.
Having the sequence of units in DNA determine the placement of nucleosomes would explain a puzzling feature of transcription factors, the proteins that activate genes. The transcription factors recognize short sequences of DNA, about six to eight units in length, which lie just in front of the gene to be transcribed.
But these short sequences occur so often in the DNA that the transcription factors, it seemed, must often bind to the wrong ones. Dr. Segal, a computational biologist, believes that the wrong sites are in fact inaccessible because they lie in the part of the DNA wrapped around a nucleosome. The transcription factors can only see sites in the naked DNA that lies between two nucleosomes.
The nucleosomes frequently move around, letting the DNA float free when a gene has to be transcribed. Given this constant flux, Dr. Segal said he was surprised they could predict as many as half of the preferred nucleosome positions. But having broken the code, We think that for the first time we have a real quantitative handle on exploring how the nucleosomes and other proteins interact to control the DNA, he said.
…
The nucleosome is made up of proteins known as histones, which are among the most highly conserved in evolution, meaning that they change very little from one species to another. A histone of peas and cows differs in just 2 of its 102 amino acid units. The conservation is usually attributed to the precise fit required between the histones and the DNA wound around them. But another reason, Dr. Segal suggested, could be that any change would interfere with the nucleosomes ability to find their assigned positions on the DNA.
In the genetic code, sets of three DNA units specify various kinds of amino acid, the units of proteins. A curious feature of the code is that it is redundant, meaning that a given amino acid can be defined by any of several different triplets. Biologists have long speculated that the redundancy may have been designed so as to coexist with some other kind of code, and this, Dr. Segal said, could be the nucleosome code.
fascinating…But I don’t really understand what the nucleosomes are doing. When there is some code responsible for low mutation rate in some part of the genom, then this code should be common to all preserved sequences in the database. Maybe we can find it ?!?
Well, it seems perfectly clear to me that the variation in HA is due to recombination, with a tiny component of actual cosmic-ray-from-outer-space mutation. The base rate of cosmic-ray-mutation is seen in PA. And this three-dimensional structure parity checking is part of the reason. My feeling is that the overall pH quality of the triplet—that is, the sum of the acidity or basicness of the triplet—is what the nucleosome ‘feels’. If the triplet doesn’t ‘feel’ right, either in shape or in pH, it gets rejected. It seems to me that looking for where on the structure the invariant sequences are expressed to see if they are adjacent to the nuclesome side or the outside would tell you. Is there a good way to do that?
wetDirt:
You can; just less of it is all.
Purifying selection (usually referred to as “stabilizing selection”) doesn’t require magic; it’s a matter of simple logic. The fitness landscape always features more ‘valley’ than ‘peak’ — a LOT more. Therefore, beginning from a point anywhere up the slope of a fitness peak, a leap in a random direction is much more likely to be a trip downhill than up — and the larger the leap, the less likely it is to be an improvement. This concept is at the heart of the “Selectionism versus Neutralism” discussion Monotreme and I had some pages back, and it’s also the argument against Goldschmidt’s “hopeful monsters”. Evolution is the accumulation of small changes.
Things get convoluted when we begin to consider mutability as a trait in itself. In a suitably susceptible host, high copy fidelity in an influenza virus would seem to represent a fitness peak, as it would lead to production of greater numbers of viable virions. But immune response by the host would also be facilitated by uniformity in viral antigens, so lower copy fidelity (most decidedly not the same thing as cosmic-ray-from-outer-space mutation) would represent a peak as well, enabling the virus to more easily outflank host defenses.
There is an infinite regress lurking here. If mutability is a trait, then it must itself be subject to varying degrees of mutability, and we may consider the mutability of that mutability, etc. In addition, a fundamental concept is that evolution cannot “see” the future. If it works, it’s because it works right now; organisms cannot sacrifice their own fitness for the benefit of their progeny. The solution to this puzzle seems to require invoking selection acting on viral swarms as well as on individual virions (and discussions on issues a tricky as that tend to degrade into mud-slinging sessions even among leading theorists; never mind FluWikians).
If someone wanted to posit secondary structure or PH or whatever as clues used by a highly sophisticated polymerase capable of selectively varying its fidelity, I’d much rather look at that than continue to be told how obvious it all is just from looking at the sequences.
The Weizmann/Widom article sounds interesting, haven’t looked at it yet (your link didn’t work for me, btw). Probably not very relevant to the replication of an RNA virus such as influenza, however, unless we’re ready to open yet another Pandora’s box, that being the subtleties of viral interaction with host cellular machinery.
here is a link, which works for me:
http://www.nytimes.com/2006/07/25/science/25dna.html?pagewanted=print
Speaking of the redundancy in triplets, flu uses it to full advantage and actually makes two genetic versions of the same protein. If you look at differences between closely related viruses (like the various H5N1 Qinghai strains), you will see that most chnages are third base transitions. Thus, most of the polymorphsism are really bimorphisms (T switches with C, and A switches with G).
H5N1 knows how to read. Most who sequence its genes, don’t.
Racter at 15:22 But…purifying/stabilizing selection is an assumption that is necessary because of the assumption that the polymerase is highly inaccurate. If you get rid of the assumption of inaccuracy, you get rid of the necessity of having a separate mechanism to clean up after the first, but only in certain places, which also necessitates an assumption that there are written instructions somewhere for knowing what the context is, and thus whether you are in quick’n’dirty or slow-n-careful mode. So you end up with this whole chain of assumptions, including the tunnel-of-mirrors mutability of mutability thing.
If you let the polymerase be accurate, then you use the virus swarm to store the variability information. Sure seems simpler to me, and involves far less administrative overhead.
And as Niman points out, flipping T to C and A to G doesn’t change the pH-ness very much, which wouldn’t trigger the error-correcting mechanism so much. So these little tweaks can be tolerated, and are perhaps useful when working with blasting out of duck versus cat cells.
I am convinced that if nobody looks for an error-checker, they are almost certain not to find it.
The third base differences limit recombination. There are more transversions when the flu is isoalted from different species. Flu has a very fine tuned set of rules.
It knows what it is doing.
The sequencers don’t.
wetDirt at 11:54
I’ve been working on that one for awhile also wetDirt.
Should it be gone by now?
Even with some admission from the random camp that certain areas of certain segments are more conserved than others (which, of course, is a rules-basis and not random), we would expect the so-called accumulation of mutations to de-select all but the most stable strains.
Remember that some have been arguing about the impossibility of stability in one strain over less than 30 years. General CW expectations are persistent instability.
We can mix and match ideas like SNPs, but I don’t think that they congeal in the end and the hybrid that occurs isn’t viable.
Our easy explanations just don’t match the evidence.
A rules-based, pre-emergent system of genetic acquisition is the best explanation for the evidence.
Niman and Monotreme,
Please take the list of words/topics and make a paragraph that communicates to the average reader.
You may work together or separately. There is no time limit and you may use two number two pencils and one whiteboard.
Feel free to submit drafts and a final version if you have several ideas. I’d love to see how your ideas develop(ed) on this one.
I’ll be back later tonight to study your insights.
Thank you for participating in this endeavour.
Gather and Solve.
wetDirt at 17:39
Large, promiscuous gene pool interchangeable at sub-unit levels. Yes?
I’d just be happy (momentarily; I’m sure I’ll always want more in the end) to know the following:
What does pre-emergent mean??
For ‘purifying/stabilizing’ selection thing: Where, exactly is this supposed to occur? Right at the polymerase? Inside the cytoplasm? During assembly of the viron? At the time of cell rupture? I’m having trouble with this. Has anyone seen it? Have they found up 100 virons with defects in the PA gene that all came from one cell? Or is it like legos, you can only fit the parts together rightside up? Where are all those defective progeny? I want to talk to them.
Polymerases: What shape is it? What is the geometry of the polymerase/RNA strand? Is there some kind of key/lock shape involved as the polymerase handles the RNA pieces? Other than locating the stop things.
Is the viral RNA all strung out or wadded up into an origami ball when it infects the cell? Is it chopped up or read out in a line?
Niman: I remember the figures in your patent application that show the symmetrical A/T and C/G flipping. It was pretty impressive. I get that homologous recombination works best with close viral relatives, and worse with distant ones. But I don’t see what the advantage of flipping bits is, other than sowing confusion in my head. Does it make copying faster, kinda like working lefthanded and righthanded at the same time? OR do you see this as the error-checking mechanism itself: As the strands are copied and assembled, the assembly process ‘feels’ something wrong in the third base bit, and chokes, and junks the copy and starts over? Would this apply more to the PA gene, and less to HA?
+++tries to stuff brain bits back into skull+++
oops, I really meant: flipping T to C and A to G
Folks, nucleosomes only exist in association with the DNA of eukaryotic (basically, nucleus-containing) cells. They are not present in prokaryotic cell DNA (for instance, bacteria); and most certainly not in association with any viral DNA or RNA. So you can eliminate any consideration of nucleosomes in conjunction with the RNA of influenza viruses.
If you want verification, check out this online book. About half way down the page is a section titled “Nucleosomes Are the Basic Unit of Eucaryotic Chromosome Structure”.
I’m not saying that RNA viruses use nucleosomes, only that the third position is used for error checking. The viruses use the host cell’s machinery for reproduction. If the host cell’s machinery does error checking, then the virus doesn’t need to do its own, it merely needs the required checksum code.
then, how can we have viruses which just differ in one bit, as seen in Karo ? My idea would be to specify 20–100 markers in the genom and specify the mutation rate in each segment independently.
Pre-emergent - non-random, rules-based system of genetic acquisition, deterministic and prior to assembly . . .
as opposed to so-called natural selection, an after effect acting as a moderator of the often discussed many defective progeny, post assembly.
see [[ http://www.nature.com/nature/journal/vaop/ncurrent/full/nature04979.html|here]] for a full article from March
NS1 at 23:09
as opposed to so-called natural selection, an after effect acting as a moderator of the often discussed many defective progeny, post assembly
NS1, you do believe in Natural Selection, right? This is the second time you’ve used the term “so-called” natural selection. What’s with the “so-called”?
wetDirt, I understand that you would like to apply your knowledge of programming to biology, but I’m sorry, it’s just not going to work. An engineer didn’t design viruses, or us for that matter.
As regards, purifying selection aka stabilizing selection aka negative selection, I’m not sure you’ve got the concept. Zillions of viruses are made. Many of them are defective. The defective ones don’t replicate. Therefore, we do not observe them. That is purifying selection. The mechanism of action is death or failure to reproduce. It applies to viruses, frogs and human beings.
wetDirt:
I, for one, do not find all of this to be drop-dead simple, and when I see selection referred to as an “error-correcting mechanism”, I feel better, reassured that I’m not the only one capable of getting lost. Remember that the term “natural selection” is a metaphor. Because it is such a compelling metaphor, there is a risk that its metaphorical nature may easily become rather transparent. Selection is certainly not a “mechanism” in the same sense that, say, a polymerase or a ribosome is a mechanism, and it is not a property which resides in either a genome or a phenotype. It is a natural consequence of imperfect replicators exploring a design-space offering vastly more ways to fail than to succeed.
As Mono has just pointed out, the principle behind stabilizing selection applies to any life form, and has a particularly strong influece on organisms well-adapted to stable environments (which can be confidently said of H5N1 in avian hosts, once the question of whether a virus qualifies as an “organism” is resolved). I believe it was Fisher who introduced the “microscope analogy”: starting with a microscope partly in focus, the odds would overwhelmingly disfavor any large random change bringing the image into sharper focus, whereas a sufficiently small random change would have a 50/50 chance of being an improvement. The greater the change — or the shorter the distance from the starting point to the adaptive optimum — the less likely it is for any change to be favored by selection.
In this context, we are asking whether stabilizing selection offers the best explanation for the observed homologies in sequences, and I have previously said that I am not entirely satisfied that it does — but whatever else is going on, I think a very safe assumption is that stabilizing selection is at work on this virus — indeed, it would seem to be an indispensible part of any theory resting on highly accurate polymerases, unless the theorist would wish to propose that this accuracy sprang forth fully-formed.
With a sloppy polymerase, you’d get lots of mutants, most of which wouldn’t work, and which would be eliminated by selection. No administrative overhead required. For the entire population to avoid “mutating into chaos”, it is only necessary that some of the mutants survive (or that not all are mutants). Where we see more variability in particular areas of a genome, we may posit that structural or functional constraints impose simply tighter constraints in those areas, increasing the likelihood of failure where changes are more pronounced.
Racter, I pretty much agree with everything you said at 00:30, including the part about selection not explaining all of the H5N1 sequences that have been deposited. There are definitely some anomalies.
Here’s a paper that’s apropo, but which requires a subscription, darn it.
Monotreme at 23:44
On so-called Natural Selection . . .
Racter says it well . . .
Perhaps we should call it natural consequence or failure or just death.
Otherwise, we are invoking the chief metaphor of the religion of strict Darwinism. We capitalise it as Natural Selection as if it is a mechanism that takes action, when, in fact, the metaphor describes a post-event limitation, an intransitive that should be defined as such for the unknowing eyes of many of the readers here.
Yes, I’ve always been troubled by the propaganda behind the terms that we use so flippantly.
how about this: the polymerase has several methods to copy the genes, it is possible to switch the methods - maybe it’s only some delay or detour which increases the probability of mutations. Then, once a certain subsequence of nucleotides is read, the polymerase switches modes accordingly. This triggering subsequence might only involve synonymous bits in certain regions, so not to disturb the other functions.
Racter at 00:30 “…With a sloppy polymerase, youd get lots of mutants, most of which wouldnt work, and which would be eliminated by selection. No administrative overhead required. For the entire population to avoid mutating into chaos, it is only necessary that some of the mutants survive (or that not all are mutants). Where we see more variability in particular areas of a genome, we may posit that structural or functional constraints impose simply tighter constraints in those areas, increasing the likelihood of failure where changes are more pronounced.”
I see it quite the opposite way. Where we see highly conserved areas, the constraints are tighter, and any deviations are nonviable. Where we see variable areas, the constraints are looser. To use what Monotreme would consider a bad programming example, the areas that don’t change are like the parts of an bank’s program that calculates your monthly checking balance. Any random mutation in this subroutine could have disastrous consequences, but they would be found pretty fast by the sound of screaming. However, the part of the code that formats and spaces the page is highly variable from version to version: things might be bold, italic, or underlined, but the information doesn’t change. The core functions are highly conserved, and the decorative functions are variable. Likewise, here we have a NP gene that is largely highly conserved. Why? This is the part of the code that is the nucleoprotien, that codes for reproduction itself: the payload. Where do we find the most variation? In HA, the part that has to deal with shifting antibodies. That code design makes all kinds of sense.
What I get from your last line is that where you see mutations, you see mistakes. When I see the same mutations, I see antigen-defeating shifting camoflauge.
Let’s take this a bit further. If most viral progeny are largely defective, then when the cell blows, it releases a mixture of bad and good copies. Some bad ones wander around in misery, unable to break into a cell, or are squashed by antibody response. Others do succeed in starting another cycle, but it’s the same story: a lot of progeny are messed up. Now a nice doctor gets a sample. That sample consists of a mixture of bad and good copies, since there is no way to prevent a serious number of them from being bad. The lab grows the sample on a different substrate, and the ones better-adapted to that substrate are the ones that show up in the sequence. Somebody else gets a sample of the lab’s sample and decides to make a vaccine based on it, assuming it is the same thing that was good at infecting human cells. That makes me a little nervous.
This model suggests it would be more likely than not to see a lot of variation in serial samples or duplicate samples from a single patient. Not, however what was seen in Karo.
Oh, and I found the answers to some of my questions from last night, but don’t have time to go into it right now.
wetDirt:
Right. My sentence got munged during an edit; it should have read: “Where we see less variability…”
As for the mutations/mistakes distinction, there are at least two different things happening: SNPs (single-point mutations of the sort caused by “cosmic-rays-from-outer-space), and errors during transcription. (In addition, reassortment may produce non-viable virus during final assembly). The fun part is considering whether a high rate of variants (of the second type) is an explicit design feature, and what logical difficulties that encounters upon recursion.
Since this one’s getting pretty long, I have the next installment open.