Skip to main content
Topic: peak picking error (Read 5408 times) previous topic - next topic

peak picking error

hi,

I have tried to use centWave to find peaks in my data. I used a run with a known compound in it. I know at which m/z and rt values I should find it. Plotting EIC with m/z range and time range around those values I can see a clear peak. However it is not detected by centWave. I used default parameters and ppm= 15.
Any ideas?

Paz

Re: peak picking error

Reply #1
Have a look at the following help page. The peak might not be as nice as it looks in the EIC.

http://http://metabolomics-forum.com/viewtopic.php?f=25&t=154&sid=1af1b9cff6303e147ab8e0c37fcd8626

Why my peak isn't picked
~~
H. Paul Benton
Scripps Research Institute
If you have an error with XCMS Online please send me the JOBID and submit an error via the XCMS Online contact page

Re: peak picking error

Reply #2
What type of instrument are you using ?

Re: peak picking error

Reply #3
We use a S1200 LC - Q-TOF Agilent 6520 instrument.
We first change the output format from .d to .mzXML using msconvert.
(msconvert data.d --mzXML --filter  "peakPicking true 1-" --filter "threshold absolut 100 most-intense")
Then we process it using XCMS, using centWave for peak detection using default parameters except for ppm.
We figured out that the ppm window we were using was to narrow.
When we use ppm = 15 for the detection our peak is not picked.
Using a larger ppm (ppm=30 or ppm=50) the peak is detected, however we get around 30.000 "data insertion problem" warnings. Are that many warnings acceptable?


Here are  the plots of the ROIs from the peak in question and the eic plot.
Thanks very much for your help

[attachment deleted by admin]

Re: peak picking error

Reply #4
We use ppm=30 for the 6520.
Do you still get warnings if you use 20 or 25 ppm ?

However, the feature's centroid distribution look odd. Could be a problem with the instrument or detector saturation ?
There is this sudden jump in m/z at the highest peak intensity ~1000 seconds. This causes the problem.

It also looks like it is detected at 15ppm, but then probably reported as two different features, with slightly different m/z ?

If you set verbose=TRUE, centWave will return an additional column "dppm" that indicates the m/z distribution of feature's centroids.

Re: peak picking error

Reply #5
Ralf,
Thank you very much for your help.
I think you are right, it's probably a saturation problem.