Log in

No account? Create an account

Previous Entry | Next Entry

keeping things interesting

So I have two sets of data, from two different experiments, which are trying to determine the same thing. Both sets of data have nice, tiny internal variation, and a good correlation (R-squared value). All good. The only problem is that there is a 10-fold difference between the two data sets. Not good.

I discussed these results with one of my supervisors, and we got into a long discussion/debate about what to do next. I, personally, did not trust one of the sets of data as it was using a methodology and equipment I am not well versed in (flow cytometry), but he was ademant that those results were a "correct" representation of what was going on. He wanted me to do something I know my other supervisor had argued *against* me doing in the past. When I told him I wanted to discuss the proposed method with my other supervisor first, I was told that there was no point because he (the other) would automatically discount it.

He then spent a goodly amount of time trying to get me to come over to his side of the fence, while simultaneously saying that it was my PhD and my decision, while at the same time saying that I'd have to have a really strong argument to support my decision.

I stuck by my guns. I refused to make any decision there and then. I told him I'd think about it.

I spent the rest of the day talking to the person in the department with the most experience in flow cytometry. We went over everything with a fine toothed comb, and didn't really come up with any reason why there should be a 10 fold difference. I spoke to the other person in the department who has been using the same flow cytometry method, just with different bacteria. He too was getting a 10 to 100 fold discrepency with his results, even though the results themselves were internally consistant.

The up-shot? What I was saying to my supervisor was correct; there seems to be something funny going on with the flow cytometry. I was right to be cautious about trusting the results as they stood. And, as a consequence, I may not have to do the method that my other supervisor did not want me to do. At least, not until we work out what is going on with the flow cytometry.

Which is good. I like being smart - it doesn't happen very often.


Mar. 31st, 2005 01:05 pm (UTC)
Ten fold is just nuts.

What *is* interesting is that my results are giving cell numbers in the y x 10^7/mL, and the other guy who is using flow (but different bugs) is ALSO getting y x 10^7/mL. Personally, I think either the methodology or the machine used for analysis is not doing its job.
Mar. 31st, 2005 01:17 pm (UTC)
To me tenfold is sometimes indicative of a lost/gained decimal place...or an error of +/-1 in a log scale.

Sorry I can't help more - this old-timer is all about static cytometers, manual colony counters and nephelometers.

You are right though - something is screwy with the set-up.
Yes, getting the same error in the same order of magnitude is interesting. If you figure it out I'd be interested to know what s going on.