Milling around; I'm going to do this from the press room so I can take pictures of the exhibit screen. It looks like we are going to get the software demo, as it's been flashed on the screen here while they set up.
We were asked to give sartorial descriptions. Landis is in a light suit, medium blue shirt, yellow tie;
Scott his accustomed darker blue shirt. Young has a red tie. Barnett has a darker shirt. Both are in lighter suits. Hope that makes Futterman happy. He is.
The panel is present, but the buzz is loud.
BRUNET: Mr. Campbell had a birdie in golf. Time: landis 1:40; USADA 9:46. Is there a confirmation of something about stopping in the middle of the day? Software issue?
YOUNG: Computer arrived a little after 8pm, needed the password, got after 10pm; doesn't work without the printer. Weren't able to do much with it.
JACOBS: Parts of the software work without password; manual data processing possible. Passwords were default micromass/micromass.
[ no ruling about software demo? ]
q: is there any difference between 1.67-4 and 1.67-2 for reprocessing?
q: how do you know?
a: I was present for the releases; had only to do with hardware interfaces (head amplifiers).
q: GDC 1372, what is this.
a: printout of datafile to demonstrate. These are the automatic results.
a: computer times in UK time.
q: GDC 1373, what is this?
a: same file processed on a different computer with the -2
q: are the same numbers?
OBJECTION: they weren't able to get the old version loaded?
q: why show this?
a: to show the automatic values are identical in both versions.
q: yesterday, said reprocessing can effect results,
q: and you can show this on the -2 software?
a: The OS is warp v3; first release 1987; Micromass selected it instead of DOS and windows 3.1.
q: how to you open something to work on it?
a: editor, command line, instrument control prog, offline program, and "dp" for data processing, no password needed.
q: to find data,
a: this data is something from Lawrence Berkeley Labs; don't know what it is exactly.
q: what do we see now without the FID
a: traces from three detectors, in 3 colors. Major is 44, minor 1 is 45, and minor 2 is irrelevant. Each peak is a compound, and two flat peaks for reference gas. Zoom in on a peak. See three peaks, almost the same time. Note these are rescaled, and not accurate scaling in display. The ratio is the area of the two peaks.
a: then load parameter of 2/1 in middle, the intensity. The 44 comes out first, which is why we get this characteristic swoosh.
q: If I want to look at background subtraction, what do I do?
a: set the peak start/end first. Here, peak at 700 and 755, way out.
a: at lndd the CGs were much dirtier. They were adjusting starts and end and background fit. So I'm going to turn BG correction on, with default parameters. Showing drastic movement, changed values 2 per mil.
q: if wanted to save, what would you do?
a: note a bug in the s/w, peaks moved back to auto, and value changed to -19. Could save this way, then load parameters, and get the reprocessed values and get the same number. I'll demonstrate closing the program and opening it again to make sure we are clear.
q: manual processing was observed at lndd?
a: yes, under the supervision of Dr. Botre.
q: what about the popup box to do calculations?
a: the popup shows the detector values as you move the cursor along. Shows you the values of the traces in the graph that is plotted at that X value.
a: not the only thing they did, also manual BG subtraction. Pink line is the automatic BG value; green is points picked by software If there's a contribution from something else, you are trying to remove it.
q: if not happy, what would you do?
a: what was happening at LNDD is they would remove points and add new points where they felt it would best fit. Demonstrate very wrong case, with massive changes to result ratio. Can go the other way as well. Can play all day and give any value you want, from plus or minus 50.
q: when you were at lndd you saw them do this?
a: yes, to the degree not extreme. Let me show minor change Determining how ridiculous the data point is is hard to determine. I don't know how to do it. I don't think the technicians at LNDD know how to do. This is an expensive, rather large random number generator.
q: if I wanted to check, I could save and document whe I'd done.
a: yes, simple save.
q: and if wanted to print, you could.
a: yes, file/print.
q: but after LNDD did this, they just printed.
a: yes, but didn't save. The SOP says you should not save the parameter file after printing results.
q: how hard?
a: file/save/params OK.
CROSS BY YOUNG:
You can turn off your computer and go back.
Something marked as an EX.
q: Talk about resume. never got a degree from cambridge?
a: correct. advisors from liverpool and cambridge.
[ he's trying to chew him up on resume fudging ]
q: staff scientist at LBL?
q: that wasn't your title was it?
a: title when I left.
q: when did it become your title?
a: can't say after the top of my head?
q: difference between Staff scientist and post-doc fellow?
a: not sure there is a post-doc fellow at berkeley.
q: let me show you another resume. instead of staff scientist, this resume says post-doc fellow.
q: so there is a post-doc?
a: my description.
q: looked through your pubs and paper list. Is it fair to say you've only had two peer reviewed articles published, one in 2001 on soil gases, and one in 1996 in bumblebees?
a: yes, I'm not an academic, whose job is to publish papers.
q: you resume, under Queens Univ, you said one of your jobs was to publish, but you have none?
[ Dunn looks very grim ]
Being done by Young, this is a delicate slime, not a Barnett.
q: when you talked about this picture about the magnets, you said you put your hand over your heart, and said you couldnt' trust the results
a: I said I didn't know what the results would be. My point is not that it would change the results, but that it shows a complete lack of understanding of the instrument.
q: I want to make it clear this has nothing to do with the isoprime1?
a: it has to do with the understanding of LNDD about the use of an IRMS machine.
q: reprocessing in May, who was there.
a: Me, Botre, Janine, Brenna [ who is Janine? ]
q: botre in charge?
q: you had access to a variety of data?
q: the original data?
a: I never had them, Dr. Botre had them.
q: and the 10 samples in April.
q: Also a log of samples in April.
q: in connection with reprocessing, was there anything you asked botre to do that he was unwilling to do?
q: was there anything asked the lab techs didn't do?
q: you talked about the ability to save the parameter files.
q: during reprocessing they did adjust?
q: you never asked them to save?
a: that wasn't my point, it was Botre's procedure. I assumed that they had saved param. I was astounded they were doing it manually without records.
q: but you didn't ask.
a: if they didn't know to do it, it wasn't my job to tell them.
q: for the ones analyzed in april, you had the opportunity to look at the data?
q: on the S17 samples or the April samples, through Botre, you could have asked the data be blown up.
a: yes. had free access, and Dr. Botre. Had the ability, but each file has over 100,000 data files, so it would not have been possible to look at it all.
q: not impossible to blow up m/z on all of them.
a: I have an excel file where I can show you the amount of data. It's not possible to look at everyone. It's burdensome and unreasonable to ask me to have anticipated each point.
q: we've talked about at most 10 chromatograms.
q: you've had them for 1/2 year.
a: I asked for the data files.
q: Dr. Botre let you do anything. Would looking at the EDFs, is that a snapshot before any manipulation.
q: so when you run the EDF's with auto BG, that takes out consideration of manual selection.
q: you asked that be done.
q: and they still came out 5a positive?
a: they aren't even the right peaks, the numbers are meaningless.
q: in your signed statement, you said the reprocessing...
a: let me look.
q: your signature, penalty of perjury.
q: read before you signed, page 6, section e.
a: trying to establish if the original was processed with newer sw., the numbers would be better; I don't think it says anything about reprocessing.
[ Landis looks stressed ]
Young is trying to rush Davis's reading.
q: let me ask the question. You requested the EDFs be reproduced on Masslynx software.
a: that's not what this refers to. I'm saying if the analysis had been done on the new software, the results would be different.
q: Previous page, iii, you say reprocessing on masslynx or ionvantage...
a: reads, "for instance, if..." yes.
q: so you say newer software would remove errors.
[ landis isn't sure what just happened. Young just gave up, because he wasn't getting anywhere ]
q: explain why reprocessing on new software doesn't guaranteed accurate results?
a: yes, may I pull up screenshots from reprocessing?
He wants directory tree shot at the very end of the masslynx data. One at 185 is missing;
GDC 1054 and 1055; missing pages from some copies.
[ davis taps fingers on rail waiting for this to get resolved ]
q: Not present in Jul or Aug 2006 for original processing?
q: Any way of knowing what they did?
a: No only do i now have any idea, they have no idea what they did.
q: when there in april/may, the numbers came up different.
JACOBS: I should have spoken slower. [ he's waiting for something to be found ]
CAMPBELL: I think I read in a declaration, a conversation with Aquilerra, where he said linearity should be run before each sample?
a: no, that was original agreement, but that was retracted. I asked if he thought that was reasonable, he said "no comment"; I have not seen a linearity run at LNDD, despite repeated requests.
[ gets screenshots ]
a: these are to shots from masslynx. "Databridge" software converts the OS2 format to Masslynx software. When we developed BG and peak software for masslynx, this was done from scratch. And we wrote this, but did a rush job, so it is only a partial conversion. So we cheat by copying stuff from other Masslynx software and samples, and we get all this extra stuff added to the OS2 files.
a: while it lets us look at things, the numbers are meaningless; 16/32 bit issues; they aren't compatible. I wanted to view this to show that the auto worked on Masslynx, and not on OS2.
a: why I requested what I requested. First, didn't know what I was going to find there, and asked for things I didn't know if I'd need; It was a shotgun, admitedly. Concluded when got there, didn't need much that was requested. But I did wish to see the raw integration, then the auto BG subtraction to see if there were not being done properely, then the manual correction on the OS2 ; then show on Masslynx to demonstrate that Masslynx could have done it properly, even if the result numbers are meaningless. That is why I was carefully reading before answering Mr young.
q: did reprocessing of S17 tell you anything about background?
a: that it wasn't fit for purpose. We don't know how it works. People have been saying "good enough"? It's not. Every single sample needed to be re-integrated, and every time we've seen significantly different results.
MCLAREN: Do I understand you that Masslynx is state-of-the-art?
a: not quite, but reasonably current. Dr. Brenna has saxicab that is excellent; isodat2; I could push my own but I won't, "maps"; Masslynx is one of about 5.
q: look at original result runs, at original and masslynx values, why are they similar?
a: they are quite different, not exact. I don't know what the difference is, but it's uncertain is the point; the lack of audit trail and know knowledge. There are quite significant changes there.
q: point out?
a: -.6 mil, .4 here, .5 here, .4 mil here; on the blanks, same thing, and there's a difference.
BRUNET: you have your software?
a: yes, it's necessary.
a: do you have commercial interest in Masslynx?