Just to clarify and add some detail: if I run XPress twice on the same (very large, multi-sample) pepXML file, once with proper arguments for H/L (8Da separation) and once with proper arguments for M/L(8Da separation), and then I compare the logs of the light areas determined with the two methods, PSM by PSM, I get a correlation coefficient of 0.87. The great majority of the light areas are exactly the same between the two XPress runs, but a large minority are far off-diagonal.
For this reason, I would not expect H/M ratios derived from the H and L areas from those two runs to behave similarly to H/M ratios somehow calculated from a single XPress run. On Fri, Jul 3, 2015 at 1:40 PM, Damon May <[email protected]> wrote: > Thanks for the response, Amit. Sorry, I'm not sure what you mean by > "correlation for only those ratios where all three labels (L M H) are > present". By definition, I'm only comparing ratios that were present as H/L > in one XPress run and M/L in the other (and so L M H are present in one run > or the other). Do you mean ratios where all 3 labels have PSMs? That seems > a bit stringent. > > > It's not that I expect the quantitation to suffer when I ignore the light > label. It's that, demonstrably, when I compare the heavy from a H/L XPress > run and the light from an M/L run, I get very different results from the > H/L run's ratio. Meaning that, presumably, the L quantity that's measured > is varying significantly across the two runs. For that reason, I expect the > H from the H/L run also to be difficult to compare with the M from the M/L > run. > > MaxQuant might be the better option for me. Thanks for the suggestion! > > Damon > > On Fri, Jul 3, 2015 at 8:52 AM, Amit Yadav <[email protected]> wrote: > >> Hi Damon, >> >> Have you tried checking the correlation for only those ratios where all >> three labels (L M H) are present?. The reason could be that finding any two >> pairs may have higher chances of false positives which may be distorting >> your real ratios and therefore resulting in low correlation. >> >> Since you wanted H/M, why do you expect quantitation to suffer when you >> ignore light label? >> >> I don't have idea about this mutilation trick. >> >> If possible, a triple SILAC search can be done with ProteinPilot and any >> ratios can be calculated even post search (M/L, H/L, H/M or any other >> combination) >> In freely available software solutions, MaxQuant can be used. >> >> >> >> >> >> >> On Friday, July 3, 2015 at 6:07:12 AM UTC+5:30, Damon May wrote: >>> >>> I've got triple-labeled data, heavy-medium-light K. We've discussed >>> before the possibility of supporting this type of data fully, and I know >>> that's not possible yet. >>> >>> However, I would like to compare medium with heavy (specifically, I'd >>> like to assess M/(M+H)). This is only a comparison of two quantities, but >>> I'm having a hard time figuring out how I might do it with XPress. >>> >>> I tried running XPress twice (H/L and M/L) and using the values for H >>> from one run and M from the other. That did not behave well at all, and >>> when I sanity checked by comparing H/L ratios derived from one XPress run >>> with H/L ratios derived from the H from an H/L run and the L from a M/L >>> run, the low correlation (~0.3) made me realize that I should not expect >>> this to work well. >>> >>> I considered re-running Comet, pretending that there's no light label, >>> and running XPress on the results. This would certainly "work", but I would >>> lose quantitation from all PSMs in which only the light was identified; >>> that's likely quite a lot of peptides. >>> >>> My only other idea is to mutilate the PepXML from my current (L-M-H) >>> Comet searches, dummying them up so that it looks like I ran an M-H only >>> search, and creating dummy M or H IDs for everything that's currently an L >>> ID. That's a really ugly solution, though, and I'm worried it'll fail for >>> some reason I haven't thought of. >>> >>> Can anybody think of another way to do this? >>> >>> Thanks, >>> Damon >>> >> -- >> You received this message because you are subscribed to a topic in the >> Google Groups "spctools-discuss" group. >> To unsubscribe from this topic, visit >> https://groups.google.com/d/topic/spctools-discuss/zr3k_VNUPgQ/unsubscribe >> . >> To unsubscribe from this group and all its topics, send an email to >> [email protected]. >> To post to this group, send email to [email protected]. >> Visit this group at http://groups.google.com/group/spctools-discuss. >> For more options, visit https://groups.google.com/d/optout. >> > > -- You received this message because you are subscribed to the Google Groups "spctools-discuss" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/spctools-discuss. For more options, visit https://groups.google.com/d/optout.
