Hearing - Tue: Davis Direct III

Coming back in. Don't see Davis -- oops, there he is in the witness box, leaning back. What up?

Don't think they're even going to finish direct today, at this rate, much less start cross.

---

SUH: we are trying to load the 167-2, hoping it will be done, so we'll move on and come back, should be able to fit it in.

[more]


q: describe the process technicians use when they did manual reprocessing of S17 data?
a: it was somewhat variable with individuals and within individuals. Peak ends were dragged and moved; there was look at ratio traces, and removal of points used by background subtraction, adding and deleting points, whle there was continual reanalsis of ratios.

q: what do you mean reanalyze?
a: [ lost point of question and answer, something with software options ]

DAVIS INVITED TO DRAW A PICTURE.

Picture shows a chromatogram with peaks, and shows auto start-stop marks, and background.

a: LNDD goes into menu, selects drag, moves start stop line.

q: have an impact on the isotopic values.
a: yes, 10s of mils. SOP says can have large effect.

q: show any other manual processing during the reprocessing.

BARNETT: please mark as exhibits.

Draws 3 peaks, maked, with ratio trace above

a: with BG sub enabled, a polynomial best-fit is done from 44/45 trace to remove background. Problem with Masslynx developemt was that we lost the code for the background subtraction in the Optima software.
This is due to corporate takeovers.

q: what lab staff would do is use the drag option and manually add data points to "better" reflect the background, reanalyze, and get another best-fit line. This can shift results by 20 per mil easily.

GDC 970

q: recognize?
a: f3 fraction.

q: describe the background trace on the OS2 program, and it's use
a: it there. as to its effect on the ratio, no one knows.

GDC 964

q: point out how you would use the 44/45 trace.
a: if you have peaks where mumble through the baseline, mumble At lndd, you'll see these boxes, here, here and here, maybe here. These are all data points the technicians have placed and we can't see, then they reprocessed the whole analysis.

a: by going to file/save params, they could reprocess and get the same numbers.

q: if you had your program, could you show how this works?
a: yes.

q: this last point - are you suggesting the os/2 allows you save the data from the first manual reprocessing?
a: yes.

q: so there is nothing to keep you from keeping the record?
a: no

q: so there is no reason you couldn't reproduce results on reprocessing?
a: no.

q: you could place the points the same place?
a: the software would do it for you. There would be a data file, and a method file. It's not a true audit trail, but you can see what the last change was. With Masslynx, you can save 4 per data file.

q: so when LNDD determined the results from S17, the program had a way to save points, so they could have been exactly reproduced?
a: yes.

q: never done?
a: yes.

q: when watching lndd techs reprocessing, did they do it over and over agin?
a: yes.

q: did you note them?
a: yes, here's a chart showing what I noted them doing.

q: what?
a: number of times peak end changed, new point to BG trace, BG point dragged, times reprocessed.

[ bunch of changes -- need to see chart to see the details ]

q: in your opinion, why was this being done?
a: she was struggling to get the line to fit the way she liked. When I asked what she was doing, she said she was "using her experience"

q: did this change the CIR results?
a: yes, significantly. sometimes so much they gave up and reloaded.

q: what are these?
a: The A(?) block are Frelat's. She made some setup errors, and reloaded a few times.

q: are these numbers the times they tried to adjust of correct the values to bring within the previous values?
a: I don't think they were trying to get the numbers; I think they were trying to fit lines and peaks the best they could. Don't know the motivation. Asked repeatedly what they were doing, and was answered they were using their experience.

q: were the CIRs changing.
a: yes, significant as stated in SOP.

Q: you watched Mongongu and Frelat conduct their procedures.
a: yes.

q: did you see them during retesting?
a: yes.

q: concusion about their ability and competence to operate the instruments.
a: they clearly did not understand the instrument. I had to help them load the software on the machine. They were obviously trying to help each other during the reprocessing and did not generally know how the software worked.

GDC 1356.1, wolfgang's relative retention time chart.

[ description ]

q: looking at these values, were you able to see the relative times for the A and B were substantially off?
a: yes.

q: from the perspective of someone familiar with the isoprime, does it affect your opinion of the reliability of results?
a: yes, you can't be sure they've identified the right peaks, the numbers will be meaningless.

q: as part of your experience with isoprime instruments, have you had opportunity and responsibility to review chromatography?
a: yes.

q: does it matter?
a: chromatography is about separating. If you don't separate, you have wrong results.

q: show some CG's, LNDD 1052, LNDD 1110 (EX 88) F3 from S11. Quality?
a: very poor. This is not a clean baseline. this period of the FID is not flat. This suggests something bleeding from the inlet port. Highly unusual. I see no baseline separation for the alleged internal standard.

q: do you know what the isotropic of this area here is?
a: no clue.

q: lndd 870, Lndd 894, seen this? stage, fraction?

[ dunn looks back, grim. ]

a: f3 of S15.

q: opinion of chromatography here?
a: looks pretty poor, at the printout. hard to tell without being able to zoom into it. I'd be very cautious using results from that.

q: LNDD 968 and 991. opinion?
a: the MSD looks a little better, but can't zoom in; co-eluting peaks; peak 13's tail is identified as a peak. getting a valid isotope n umber from that would be impossible.

q: [ lost page id's ]: opinion?
a: pretty poor. look at central peaks , like 11; strange integration. The sw has decided to draw line here; it's not valid; on the other side these are all co-eluting. Very difficult to say you could use numbers.

BRUNET: 5:00, will adjourn

BARNETT/YOUNG: not confirmed rebuttal witnesses.

SUH: going to call any?

YOUNG: expect to call Brenna.

ADJOURNED until 9:30 am.