Tuesday, December 23, 2008

New details from the genomic forensic analysis

Dr. Jacques Ravel of U Maryland, formerly of the Institute for Genomic Research (TIGR), spoke at the MIT faculty club on December 16, 2008 to discuss the scientific evaluation of the anthrax letters' spores.

I requested but did not receive an invitation to the event (despite my Bachelor's degree in Biology from MIT and some expertise in the subject). The slides were shared with me later, but each is labeled "do not distribute."

So I will not share them. They should be posted soon on an MIT website associated with the program that brought Ravel to MIT.

Of interest, the morphologic variations in spore colonies were not entirely identical between the NY Post and Leahy letters. Slide 16 indicates that 5 variations were found in Leahy's anthrax, and (only) 3 of those 5 variations were found in the Post anthrax. No other samples had these 3 variations but some (of greater than 1,100 samples screened) had 1 or 2.

There were no genetic polymorphisms between the 2 samples or a wild-type Ames.

Also of interest, Professor Ravel pays a lot of attention to the criminal-legal aspects of the case and the rules of evidence for trial.

An early article on the genomic studies in the Times is worth a read for comparison.

23 comments:

Anonymous said...

So, the inoculant for production of the material that was used in the attacks didn't come directly from Ivin's flask, now did it?

Damn inconvenient truths!

Hint to the F.B.I.: "Passages" is more than just the title of a self help book from the early 80s.

Meryl Nass, M.D. said...

It is impossible for me to comprehend why or how the FBI can continue to disseminate information that turns out to be dead wrong, once the facts are known or someone carefully parses their claims.

To make a big deal about 4 morphologic findings that were present in the anthrax letters--and then to learn one sample had 3 and the other 5, MEANING THEY WERE NOT IDENTICAL--defies understanding.

IS FBI simply practicing its own version of Hitler's Big Lie? As paraphrased by the OSS: "People will believe a big lie sooner than a little one; and if you repeat it frequently enough people will sooner or later believe it."

The case still isn't closed. I'll wager it doesn't close for a very long time. Simplest way to keep the rest of the evidence hidden.

Anonymous said...

From the very beginning, theinvestigation has followed
The Prime Directive:

Any Line Of Investigation Is Open.

As Long As It Doesn't Lead To Battelle.

Anonymous said...

Mega bucks spent on what we are told is a criminal investigation which produces Lies, Deceptive statements, 200 new ways to weaponized anthrax, the decoding of the anthrax genome, and the persecution of and innocent person.

That’s not a criminal investigation that is a Banned Bio-weapons Program.

Anonymous said...

One sample having 5 and the other 3 must mean that the first had all 4 of the morphological varients plus the orginal (for a total of 5), while the second had only 2 plus the original (for a total of 3).

So one had all four, but the other only 2.

I wonder what answer Majidi has for this.

Anonymous said...

I think I know how Majidi would answer - he would use the M&M analogy he already made to a scientifically illiterate media. Majidi claimed at the briefing that drawing a small aliquot of liquid spores from RMR-1029 was just like taking a handful of M&Ms from a bag full of different colored M&Ms - you may not draw all the colors in your handful.

There is a lot wrong with Dr. Majidi’s analogy with M&Ms - but then again Dr Majidi’s job here is to pull the wool over the media’s eyes - not to make scientific statements.

What are the odds that, in the last breath you just exhaled, that breath contained at least one atom from the last final breath of a the dying Julius Caesar?
That’s a common question asked of science undergraduates to help them understand the statistics of very large numbers. Avagadro’s number is very large - thus the chances that you just exhaled one of Caesar’s last oxygen atoms are near 100%.

A similar situation applies to drawing a small sample of spores from the master RMR-1029 flask at Detrick. Because there are trillions of spores in the flask, you will draw a statistically significant sample of them - even if there are less than 1% mutant variations, you will pick up a significant number of them. This ain’t a bag of M&Ms where there are only a few hundred in the bag and you draw out a handful.

If I understand the new information correctly - that there are now FIVE variations in the Leahy anthrax instead of just the FOUR in the RMR-1029 flask, it means either that, by chance, a NEW mutant was made during the creation of the Leahy batch - or else there is ANOTHER master flask somewhere (Battelle anyone?) that contains all FIVE. It also means that the NYP material was taken from an original DIFFERENT flask (not RMR-1029).

Anonymous said...

Remember also that Battelle had a project aimed at creating mutations in B. anthracis to emulate the Soviet drug resistant strain.

Anonymous said...

The proposed Battelle project was to duplicate Soviet research where it was reported that a strain of b. anthracis was genetically engineered to express an additional toxin, cerolysin, which is normally produced by bacillus cereus. Since the Soviet anthrax vaccine is essentially the same as ours (confers immunity against the two toxins produced by b. anthracis) their genetically engineered bug was resistant to their vaccine (the vaccine conferred no immunity to the new toxin their research strain produced). It wasn't really a "mutant" (rather, it was recombinant), and
"vaccine resistant" normally isn't the same as "drug resistant" (usually refers to antibiotic resistance).

As far as I know, this project never went forward. Battelle was however in charge of the late-stage development of Ivin's rpa-102 next generation anthrax vaccine, from either pre-clinicals or phase I throught phase III.

Did Majidi actually use the M&M analogy? He said this with a straight face? He really can't be that stupid, what is so amazing is that he thinks everyone else is.

What a jerk!

Anonymous said...

"Did Magidi actually use the analogy"?


Oh yes he did....

http://anthraxvaccine.blogspot.com/2008/08/papers-of-record-unsatisfied.html

Majidi
said it was like a bowl of blue M&Ms that had mixed in it a few
that were brown or green or red. The fact that those were in there was
like a fingerprint — potentially a way to trace the anthrax in the
letters back to its source.

-----------------------

Hey guys, this is FBI science we're talking about here, not real science. The FBI invented their own laws of science a long time ago:
http://www.cbsnews.com/stories/2007/11/16/60minutes/main3512453.shtml

They "invented" this lead in bullets "science" that nobody questioned until someone actually did....

Let's hope NAS gives them oversight for their anthrax "theories", because this is the same ol' same ol' FBI science labs making up shit as they go along.....

Anonymous said...

The FBI is going to have to explain this. Either to the American public sometime in the next few weeks, or before congress during next year's congressional hearings (which at this point I would say are an absolute certainty).

There are just too many unanswered questions for this to (like they were hoping) quietly go away. And who knows what else will be revealed in the weeks to come.

Anonymous said...

Anonymous wrote:
"The FBI is going to have to explain this."


I would hope that they can. If anyone actually watched the 60 minutes video I linked to above, they may have noted a very embarrassed FBI labs director Dwight Adams admitting to the lead in bullets FBI labs fiasco. That was eventually overseen by NAS as well. Let's hope NAS still have some sharp folks ;)

Does anyone get a feeling of veja vous watching this video and then listening to Dwight Adams's latest comments below on the anthrax case? Do you remember the Who's song, "We won't get fooled again?"

http://www.nytimes.com/2008/08/16/washington/16anthrax.html?pagewanted=print

Dwight Adams, a former director of the F.B.I. laboratory who was deeply involved in managing the anthrax genetic research until he left the bureau in 2006, said he was confident that the groundbreaking forensic effort would be validated by the broad scientific community.

Recalling the early skepticism that a genetic fingerprint of an anthrax could ever be obtained, Mr. Adams said, “I think the bureau and the national assets, including the national labs and others, that were applied as a team can very easily defend what they did and the results.”

Old Atlantic Lighthouse said...

To avoid confusion, the FBI or someone has to give names to each morph and list in a table which ones were in which letters.

Its possible that there is also not a unified definition of what 4 variants means in the project. Some people think it means some sort of original plus 4 additional and others think that 4 means the 4 named morphs that are distinct from the original.

They should also have a photo of each colony and its name and explain why they think they look different from other morphs and what the variation of appearance is when the same morph is present.

Also, what are the percentages of each morph in the original flask and in the other 7 flasks (or is it 8?) that have it. Give the name of each lab that has the same batch as at Detrick.

Also what about other morphs they decided not to pursue? Science is supposed to falsify hypotheses not just convict dead men. How about a second team tries to break the first one's results by finding other morphs not in the Ft. Detrick flask but in someone else's?

Anonymous said...

Also, explain if the process of isolating the b. anthracis from the b. subtilis from the media material could have influenced the ratios of morphs.

Any microbiologist with even a little experience will puzzle over this one.

Old Atlantic Lighthouse said...

What about a single morph in one letter that is only in one lab flask? Or is in a lab flask different than RMR1029 and not in RMR1029? Those cases should be pursued next. Basically, any morph in any letter is back tested against all the flasks. Publish the entire table.

Old Atlantic Lighthouse said...

The FBI's thinking is: We know in advance the letter anthrax all came from a single flask, that of the lone nut. They decided that in advance, and their methodology builds that in. However, if one drops that assumption, there are other possibilities. Two or more sources sent independently. Two or more with mixing after powder is produced, possibly because not enough, and so on.

Old Atlantic Lighthouse said...

"he spread out some attack spores on a bed of nutrient and let each form its own colony. All the colonies looked identical except one, which, to his trained eye, seemed very slightly different. Different-looking colonies are called morphotypes or just “morphs.”"

You can't tell until you grow them if they are different. So the anthrax in the letters can contain any number of morphs. Unless you grow all the anthrax in a letter you can't know if you got all the morphs in the letter. So we have to have a stopping rule. One stopping rule is do all. But they probably kept some.

"Dr. Ravel was asked to decode seven more morph genomes,"

So 8 morphs total.

But there could be more.

The entire procedure had a stopping rule of convicting a lone wolf scientist. Stop growing and testing and comparing when you do that. But that is not trying to disprove your hypothesis.

Old Atlantic Lighthouse said...

If you take anthrax powder and spread it out and grow it, isn't it random which spores produce colonies? So some morphs in the sample don't grow to colonies and some do? So it requires a statistical test? Looks like the entire procedure needs to be tested by going forward and then backward under controlled conditions. Also start with single flask or with multiple flasks with different morphs or overlapping morphs and see what happens.

Old Atlantic Lighthouse said...

Each letter had about 1 trillion spores. So they didn't grow 1 trillion colonies, they grew say 10 to 100 or 1000.

Suppose out of the 8 morphs mentioned at one point, each is present in a sample (flask or letter) in the amount of one percent each. If you grow 100 colonies from that flask or letter, then for those morphs you have a 1 in 100 chance to get it.

This applies to letters and to lab flasks. They had 1000 labs they tested? 8 had the 4 morphs they were checking for? If all 1000 labs had all 8 morphs but each lab grew 100 colonies per lab, then how many labs would have the 4 show up?

What if some labs grew 50 colonies? What is some grew 10? What if the Ivins flask, they grew 10,000 colonies and the other labs they grew 100 colonies?

There are quants from Lehman with nothing to do, why don't they hire them to do Monte Carlo.

Old Atlantic Lighthouse said...

Maybe the 8 labs that found all 4 morphs were the 8 labs that grew 1000 or more colonies? The other labs grew fewer colonies? Monte Carlo was used at Los Alamos early on. Vahid Majidi is from Los Alamos.

We need a table of how many colonies grown by each lab and which morphs they found. We also need to know how many times they went back to grow more colonies from the Ivins flask until they got all 4 from that flask they were looking for, and whether they decided which of the 8 to look for after they concluded 4 were not in the Ivins flask.

Old Atlantic Lighthouse said...

"Though 99 percent of its spores were identical with the Ames ancestor, some 1 percent or less were morphs. "

" As the morphs became available, the F.B.I. started testing samples. At first, some had one or two of the morphs. None had three of the morphs.

By late 2005 to 2006 it became clear that just eight of the 1,070 samples collected included all four morphs."

Because they started growing more colonies when they didn't find all 4?

Let p[i][j] be the percentage of morph i in flask j. This is not a constant of nature. A flask can have a low percentage but still be there. If one grows 1000 colonies and p[i][j] is .01 in one flask and
.0001 in another, then you may not get it when testing that flask.

If a person was the bad guy he could just grow fewer colonies from his flask and he would not get all 4.

What about the Ivins sample that didn't test out? Which morphs if any did it have, how many colonies grown, etc.

How long have these morphs existed? How often do they form on their own? The same morphs are formed easily all the time? They all come from a single Ames lab source from 1981? Or other wild Ames has been added to lab stocks which had the morphs from hundreds or thousands of years ago?

If all lab Ames is from a single 1981 source, then wild Ames should have lots of different morphs. That could be tested.

Old Atlantic Lighthouse said...

If they grow 1000 colonies from a lab flask, that is more work to identify the morph ones. There must be random variation in the appearance. Different people are doing this work.

From NYT

"Then an Army microbiologist from Fort Detrick made an unexpected discovery. Using an old-fashioned microbiological technique, he spread out some attack spores on a bed of nutrient and let each form its own colony. All the colonies looked identical except one, which, to his trained eye, seemed very slightly different. Different-looking colonies are called morphotypes or just “morphs.”

“Had that task been assigned to someone less experienced, these morphotypes might never have been seen or their significance never realized,” Dr. Fraser-Liggett said."

So out of the 1000 labs tested, some had untrained eyes looking at the colonies grown. If each lab grew over 1000 colonies that is a problem for untrained eyes. Also compartmentalization made it harder to achieve consistency. People would need to be trained what to look for.

So if they grow 1000's of colonies per lab flask, then untrained eyes will glaze over to miss the perhaps single morph colony to be found. If they grow under 100 colonies they miss growing the morph colonies because the article says they are under 1 percent each.

Old Atlantic Lighthouse said...

Suppose taking into account all factors, the percentage in the flask, the colonies grown, error in seeing the few morphs, the chance is .3 per morph per lab flask to see a morph. Then to see 4, one takes
.3^4 which is .0081. So out of 1000 labs, one finds all 4 8 times. In this case, all of the labs had the 4 morphs just only 8 detected them.

The total of all 4 morphs from the NYT article appears to be under 1 percent, so .25 percent each. This might work out to the .3 chance to see a morph depending on the number of colonies grown and accuracy to identify the few morphs.

Old Atlantic Lighthouse said...

If 8 morphs had evolved from a single strain from 1981 and the 1981 had one only it doesn't make sense. If 8 evolve in 20 years to 2001, then the 1981 Ames should have had many many indels already each producing different colonies. I.e. evolution should have produced more than 1 indel in some sample taken in 1981.

It makes more sense that the 8 were all there in 1981 and that the chance to see each morph and record it correctly is .3 per morph per lab flask, so that 8 were seen randomly. The letters were handled with more care and more colonies grown perhaps, so they identified those somewhat more accurately despite small fractions of the indels.