Thursday, May 17, 2007

Hearing - Thurs Frelat II

Our 15 minute recess is looking more like 30; much milling, but no parties. They were going off to discuss timing of the rest of the hearing, and Bill thought that sounded like a half hour or 45 minutes.

The buzz in the room goes up and down in waves, everytime the door on the left of the bench opens and someone walks in. Here's Young.

Barnett is already here, as is Cynthia Henson, the UCLA scientist who is sitting in front of the bar, and has been acting as the USADA science advisor, passing notes up all the time. Some have questioned whether that's appropriate, but I think it is totally correct. UCLA helps USADA prosecute cases, they are the national experts.

Team Landis comes in from the right door that leads to the "jury room". Still photogs are banging away as Landis is standing up, facing this direction in his all black suit, shirt and tie. Johnny Cash? His voice is too high, I think, but the walk isn't bad.

Arnie has been reading these posts during the session. Hi Arnie!

Frelot comes in, and gives a smile to her compatriots before sitting in the box. McLaren and Campbell come in together, and flip through some exhibits. It's louder, the cameramen stand up. The one in the back right takes naps when he can.

Brunet walks in. Quiet.

BRUNET: Welcom back, continue with the cross examination of Ms. Frelot


Q: we were looking at USADA 331, the sequence file. 5 stability runs, 3 mix cal irms, one mixcal a, the blanks and the samples interleaved, and a mix cal ac.?
a: yes.

q: you were taught this needs to be run in this orders.
a: [shrugs] yes.

q: one right after the other?
a: what do you mean one after the other?

q: in this sequence one after another
a: yes.

q: USADA 360, you recognize this as the results sheet with acq 12:21:14 and current 13:45:11.
a: yes.

q: describe what the acq time and current time are
a: acq is when you do the injection, and the current is where I have verified the data.

q: at the time you did this B sample test, you said you did not know this was Landis' sample?


a: once the analysis was done, the name had already appeared in the press.

q: do you remember when Mr. Landis' name appeared in the press with an AAF from LNDD?
a: no.

q: so you're saying when you did this test, you knew that the irms test was being done was Landis'?
a: yes, the B, I did know that.

q: USADA 347, the data page for the F3 fraction, which on 331 is the one at sequence 10 just after 9, the mix cal acetate.
a: yes that is the injection that comes right after the mix cal acetate.

q: [compareing the data sheets] This one is supposed to immediately precede this one here.
a: they come one after another.

q: can you tell me the time between the form the QC sample to the blank F3?
a: you'd have to refer to the preparatory sheet for the sample.

q: how much time separates the blank f3 from the mix cal acetate.
a: about 4-1/2 hours.

q: what happened in that 4 hours and 40 minutes between the mix cal acetate and the blank f3?
a: i'm referring to the preparatory sheet USADA 0299, that's the start, and the next page also USADA 300.

q: what about this document tells you what happened in the time gap?
a: [ looks at pages, head down, with laser ] The mix cal acetate was injected at 12:24, and at that time i was finishing preparation to launch the gcms, and then on the following page, [ puzzled] the injections on the gcms started at 13:22, and then waited until all fractions analysed in the gcms, and then i could begin my analysis on the irms. The time I actually started the irms is on the next page. I placed my samples on the autosampler of the irms.

q: so you ran the calibrators even before the samples were ready.
a: yes, but there's no problem with that.

THIS IS A SUMMARY FROM USADA's BRIEF about results from vaious origina/iso1/manual/zero/masslynx data.

q: You participated in the reprocessing in the data removed from the isoprime1 under the supervision of Dr. Botre.
a: yes.

q: you helped in the reprocessing of the B and the blank for the B on stage 17.
a: yes.

q: these values on the charts for the reprocessing.
a: yes.

q: looking at these values, -2.02/-3.51/-2.45/-6.39, original; blank -1.08/-0.08/-0.67/-1.60. Turning to processing on the original sample. When you did the first analysis did you use the OS2 automatic background subtraction?
a: when a report was drawn in the OS2 software, I would always verify the estimate of the background noise and the integration of the peaks.

q: manually?
a: yes.

q: when you do it manually, you are looking at a monitor, you are choosing where the peaks begin and end.
a: of signal 2 over 1 referring to the traceability 44/45.

q: you make the determination where the peak begins and ends.
a: yes.

q: your SOP says you have to exercise great care in picking the start and stop because it can have a dramatic impact on the isotropic value of the result.
a: yes. it is written we must be careful, but it doesn't say it can have dramatic consequences.

[digs in book ]

a: It is written "one must be careful because it can have a significant change in results", not dramatic.

q: what do you believe a significant change would be in your final mil value for a target isotope, any context, as it relates to the SOP?
a: a difference of 1.5 1.6 per mil.

q: to your mind is 1.5 to 1.6 per mil.
a: yes.

q: you are aware lndd would report AAF whenevere there is a -3 per mil difference.
a: yes.

q: what is the uncertainty at lndd
a: 0.8/mil

q: in your mind, significant is 1.5 to 1.6. can you point to an SOP that says a significant difference is 1.5 or 1.6?

[ looking ]

a: it's not written, it's my opinion.


q: back to the chart, manual reprocessing -0.35/-1.61/-3.05/-7.19 done the same way as you used on the S 17 B sample originally.
a: yes.

q: you were trying to do the best job you could on this and the origina.
a: yes.

q: when you look at the results are any of them the same?
a: the delta-deltas for the 5b and 5a are close, for the others they are different.

q: you are saying in your mind, the ones that are close, are close enough?
a: yes.

q: certainly within you own judgement of 1.5 to 1.6 per mil.
a: yes.

q: when you were doing the manual reprocessing, did you maintain any record?
a: there is a sheet with results called the analysis report on two pages, USADA 351, and 352.

q: is it your testimony those pages document your choices of where the peaks begin and end when you did the S17 original B.
a: these are the results I found.

q: is there a report how you came to that conclusion when you were using your own judgment about where the peaks begin and end?
a: we do not priint any reports that show the 44/45 trace we only have the final report after we've done the work and the chromatogram with the greatest ion measure.

q: so you have a report on the results, not the process.
a: yes.

q: this process where you mark the begin and end is to adjust the background, for background subtraction?
a: when we verify our data, we verify the integration of the peak and we verify the estimate of the background.

q: background subtration is also done manually?
a: when it is not properly done, yes.

q: in this case do you recall if you did it manually?
a: may I look?

[ looks ]

q: when you say you do manuall background subtraction, you were taught to do that?
a: yes, when it is necessary.

q: and the determination of when necessary is your own judgment?
a: we have to correct the background noise when the software has placed the background on a peak.

[answering hanging question on manual]

a: I did the verification and reprocessed it.

q: For stage 17 B's when you got the original values, did you use manual background subtraction.
a: yes.

q: let me show you a question that was posed to LNDD

"8. Please explain, with mathematical formulas, how LNDD performed and applied backgorund subtraction to sammple 994574 and related controls."

[ this is from discovery request ]

The answer received:

"See response to C10." and C10 is:

"Background subtraction is embedded in the instrument software, which is proprietary to the instrument manufacturer. LNDD has no separate documentation"

Q: how was background subtraction done for this sample and the answer from LNDD was it was done by the software, and this answer is not correct.

a: well i wouldn't say that.

[ brunet's chin in hand ]

a:: it's complicated to explain. the subtraction of the background noise is done by the software and I am able to correct it if the evaluation is done on or in a peak.

q: that description is not in the answer we were given, is our answer there?
a: no.


q: lets's look at the blanks orig values for the 5aA shows manual and auto -1.60 and -3.45. and the value jumps up larger than you personal uncertainty 1.5 1.6?




q: you said a significant difference is 1.5 to 1.6.
a: yes.

q: this difference is greater than 1.5 to 1.6?
a: yes.

q: how would you explain how when you did your manual processing that you arrived at such a different value as when the automatic subtraction was done?
a: the correction or adjustment is not neccessarily right, let me start again.


[ the 5aA are -1.6, -3.45, -1.89, -1.24 and -3.66 for origina, auto, reproc manual, zero, and masslynx ]

a: the automatic is not necessarily correct. and I must have had to correct it.

[ note the masslynix ONLY has automatic, so tests run on it can't be "cleaned up" by manual ]

q: when look at the automatic correction, that result is in excess of -3/mil, correct?
a: I think it is greater than 3/mil. When I verify my data, I don't always look at the page that comes out of the printer. I look at the screen ion signal for the greatest and the signal of the 2 over 1.

q: and when you saw over 3 per mil is when you report an AAF.
a: well in the lab what you call an AAF when it is > 3.8.

[ Landis is going to say the blanks really were -3.45, not -1.6 ]

q: is LNDD required to account for uncertainty in the measurement for IRMS delta values?
a: it is not required to, but we do that.

q: not required, but do any way?
a: yes.

q: who taught you that?
a: i don't remember, but I know that is what I do.

q: what did you see when you first looked at the monitor that caused you to make your manual adjustment to get a "more accurate" value?

a: the estimate of the start and end of the peak were perhaps not correct, or perhaps there was some background noise in the ascent of the peak.

q: one of those two things?
a: yes.

q: but you don't remember?
a: no

q: and there is no record which it was?
a: no.

q: and your testimony is that you were better able to do the background than the computer automatic, so it needed to be fixed?
a: yes, because it needed to be corrected.



Q: GDC 00970.

[ this is a reporocessing from masslynx with a chormatogran and the 44/45 trace ]

q: when you look at the background drawn by the instrument automatic, with auto start/stops, do you have the ion trace in front of you, like this?

a: yes.

q: do you have the peaks like this?
a: yes.

q: at that point (pdiol) are there any numbers on those peaks or delta values?
a: yes.

q: when you go to start or stop a peak, what do you use to decide the start of stop?
a: I do a zoom, on the 2/1 trace, and on the ion.

q: is there a set of numbers anywhere that helps with that process?
a: when I do the zoom, I no longer see the delta's because they are outside the area I can see.

q: is there a set of numbers on the 2/1 that you use?
a: I only see the time scale.

q: is there a box that pops up?
a: when I do the zoom there is no box. I have to call it up. then when I bring up the box I can move my cursor over. when I want to use my cursor to move, then the box comes up.

q: how do you use the box?
a: I look to see when the 2/1 increases and that means I have a peak that is starting.

[ that wasn't the procedure Brenna said, I don't think ]

q: when you manually adjust the background. Do you yourself do the calculation of what the results are to the delta values or does the instrument?


q: when you select the peak does the instrument do the background subtraction?
a: yes.


q: when you start a peak and end a peak, is it the instrument that calculates the area under the peak.
a: I don't know any more.

q: USADA 254, bottom table [chain of possession] ; you are 26?
a: yes.

q: 11:03 right.
a: yes

q: USADA 299, blow up section a little way down. what is it?
a: [third box on line. ] it's the time I took the test sample.

q: does it correspond to 254?
a: yes.

q: i understood your testimony that LNDD does not have a single document that reflects each transfer in the chain of custody, correct?
a: I had understood the gentleman was talking about transfers between persons.

q: does Lndd have a document that reflect who was in posession of the sample when and for what purpose?
a: well the preparation report.

q: the report?
a: yes.

[ he is trying to save the CoC. ]

q: Talkingn about the April testing, and asked a lot of questions about why you could remember very specific events during that time. Was there anything memorable about that week, like the presence of three experts.

OBJECTION: leading the witness.

BRUNET: you are leading the witness.

a: that particular week of april 2007 we worked 12 days in a row and there were people there who were observing every single thing we did and the days were very long indeed.

q: when you do irms analysis and you do the instrument verification on the machine, and you determine there is something not operating properly, is it normal for you, before you proceed with the next step...

OBJECTION leading and compound.

BRANNON: are we going to get into comound objections?

BRUNET: agree with Mr. SUh

Q; what do you do when you determine a problem exists.

a: during the verification you stop the instrument and you correct it if there is an error to be corrected for the verification.

[ do they restart verification? ]


q: you testified when you were looking at the monitor doing manual adjustments, you have the ability to zoom in and look at the peak.
a: yes.

q: you need to be able to look in and zoom to make the determinations.
a: yes.

q: show me in the printout where the zoom ins are?
a: there are no printouts of the zoom.

q: there is no record of the zooms you used as the basis of your determination where the peaks should begin and end.
a: yes.

q: when you do zoom in, you said you can't see the isotropic value.
a: yes.

q: say we have a peak like this, and you zoom in, you don't see values, and when you zoom out, you see the values again.
a: yes. What's the point?

[ THAT she sees the isotropic values that result from her corrections, so when she sees a blank that is -3.4, she can correct it to a reasonable looking -1.45. ]




James said...

if Arnie is reading the posts then he knows I think he and Floyd should take you and dave on a ride sometime...


bill hue said...

I think riding is the last thing on Floyd's mind right now but TBV and Bill appreciate your comments and kind wishes!!!!
Bill Hue

Anonymous said...

Can you give us your sense as to how much, if any, her testimony is helping Suh right now? I would suspect that it's damaging to admit you use your own opinion on these things, but I'm not sure...

Illinoisfrank said...

"q: so you're saying when you did this test, you knew that the irms test was being done was Landis'?
a: yes, the B, I did know that."

She knew the 'B' sample belonged to Landis!?!? Did she know the values for the 'A' sample at that time? Was this one of the runs where the sample file was overwritten several times? How much variance can be introduced by using manual subtraction (am I using proper terms here?)? Did you leak these results to L'Equipe?

Thom said...

So she made a manual adjustment but didn't document why. So it is simply blind trust that she made the right adjustment. Interesting. That wouldn't cut it for me.

Anonymous said...

This is why manually drawing baselines on peaks is not allowed where I work.

Anonymous said...

You can draw the peak to get the desired result.

Anonymous said...

This should be damaging to USADA/WADA. They tout infallible science and now their equipment operator is testifying that as a matter of course she jiggers the computer output when she judges it to be necessary and the SOP does not reflect any of it. No data/printout of the pre-adjustment readings, or the corrections. No notation of it being done. All you are left with is the adjusted data. AND she just knows when this needs to be done.

Can't buy science much better than that if you are a prosecutor.

James said...

Ok so she can change the base line from...

- 1


- 4
- 3
- 2
- 1

and its safe to assume she knew the A sample value and admitted she knew it was Landis' B sample she was working on...

So thats it right? Do we even need to continue? I mean I'm sure we will but this is a little bit ridiculous. They shot holes in the A the other day and now they're making manual adjustments to get the B in sync...

Even if the A was legit this throws it out the window.

I'll bet Floyd wants to ride now...


Anonymous said...

Please read very carefully. Everything in this hearing is biased. Every single question is very accurate, but the ONLY GOAL is to bring together some facts but NEVER tell you if they are relevant for the analysis expertise.
The point is that none of the Landis team points are relevant. And they make YOU to bring together facts without relevance and make YOU imagine the conclusion. This is a very common manipulation. Don't catch anything that have not been explicitely said in the hearing or you will be wrong every time, unless you are a (real, not claimed) expert of the IRMS.

LNDD is one of the 2 or 3 better lab in the world, that's proven for several years. And this never seen before Panel decision will prove it again.
Landis has been doped, sorry for you.