Extracting peak picking information May 06, 2015, 12:36:51 PM "Executive version": How do I extract the figures produced by the centwave algorithm to describe peak fitting, without watching all 60k ROIs flash past?Hello xcms gurus,Having previously used xcms as a way to automate extraction of EIC data, I'm having a crack at using it to pick peaks for our systems. The problem here is that I am a little bit of a fraud here - I am not doing metabolomics, but combinatorial 'systems' chemistry, and I fear that the systems I'm looking at are less amenable to this - the result of a messy "combinatorial explosion" (ca. 60k ROI per sample; lots of samples, and this is the first stage in my complexity progression).... I am using the centwave algorithm, and a lot of the traces you get as it picks peaks have the kind of problem illustrated here:[attachment=0:3u05twde]multiple peaks to pick.png[/attachment:3u05twde]To get this image, I had to run something like the line below, and wait endlessly.Quotexset <- xcmsSet(datafiles, method="centWave", ppm=15, peakwidth=c(0.2,5), snthr=6, mzdiff=0.005, integrate=1, noise=0, fitgauss=TRUE, sleep=0.001)[n.b. settings slightly different to produce the image - different peakwidth, no fitgauss]How do I extract these figures AFTER the process (or to files during it)? That would let me supervise, and return to the same EICs time after time, to check the optimisation. Alternatively, is there another way to plot what has been picked against the real data?? The products we're looking at are peptides. In the end I hope to look at sequence bias. Any suggestions? I feel like I'm getting somewhere (slowly), but would appreciate knowing if there is a smarter/established way to do this... e.g. I can produce (v. long) lists of possible products - should I be using a more targeted approach?Cheers,Andy[attachment deleted by admin] Quote Selected
Re: Extracting peak picking information Reply #1 – May 12, 2015, 07:45:01 AM For the moment, I'm using this "pdf(fnamepdf, onefile=TRUE)" before the xcmsSet construction, then "graphics.off()" after. That does the trick, but takes aeons.Any tips from xcms ninjas for optimising peak picking for these kinds of 'busy" chromatograms? Help would be much appreciated....Andy Quote Selected
Re: Extracting peak picking information Reply #2 – September 22, 2015, 06:55:36 AM No suggestions???I'm not any further forward, and as I have tens of thousands of peaks, making a pdf of all of them is not practical (and ENORMOUSLY slow)...... Quote Selected
Topic: Extracting peak picking information