Hello,
Job ID: 103677
I am trying to look at a particular mass that I know is in raw chromatograms but for some reason cannot extract it with the parameters I am putting in.
I am able to extract it using Marker Lynx software, however.
Parameters I am setting are as follows:
Feature Detection
method centwave
ppm 25
minimum peak width 5
maximum peak width 12
signal/noise threshold 10
noise filter 0
mxdiff -0.005
prefilter peaks 1
prefilter intensity 30
integration method 1
Retention Time Correction
Method obiwarp
profStep 0.1
Alignment
mzwid 0.01
minfrac 0.1
bw 10
max 100
minsamp 1
I was following somebody's parameters who was able to extract the particular ion in question, but in a different dataset and different biospecimen.
I am unsure of which parameters I should try changing to get this mass of interest extracted.
Thank you for your help in advance!
Sincerely,
Majda
majah,
I'm going to assume that you're using UPLC rather than HPLC as you said you're using Marker Lynx. For most UPLC datasets I've seen I set my parameters as peakwidth=c(3,20) and depending on what instrument you're on the mzwid of 0.01 is a little low. It might be worth checking the raw peak list to see if the peak is there before grouping and in how many files. Using something like the following code:
## assuming your object is called xs, i.e. xs<-xcmsSet(....)
head(xs@peaks)
idx<-which(xs@peaks[,"mz"] > 264.14 & xs@peaks[,"mz"] <= 264.10 & xs@peaks[,"rt"] > (0.28*60) & xs@peaks[,"rt"] < (0.3)*60)
length(idx)
xs@peaks[idx,] ## this will show you how many files the peak was found in and if it was found.
If it finds the peak, which it should then you want to look at your groups/boxes manually to check you're not grouping 2 or more features into the same box (i.e. sleep >0). If this does start to happen I would check out the nearest method for the grouping.
?group.nearest
gxs<-group(xs, method="nearest")
Should do the trick :) Hope it helps and let us know how you get on.
Paul
Hi, Paul,
I have never used R for XCMS. I have always used the online version, a lot more user friendly for someone like me (not a bioinformaticist :)). While I can try to extract my peak manually using R, I wanted to see if there is something I can change in the online version of XCMS that can help me extract my peak.
I have inspected raw chormatograms visually and it is definitely there, although not very highly abundant. It is present in about 60% of the samples.
Can you help me change the parameters online before I move into trying it in R?
Thanks so much!
Majda
And yes, I am using UPLC, not HPLC.
majah,
I'm so sorry I didn't see where you had posted I just followed the RSS feed to your question, my fault.
If your intensity is low it might be worth seeing if you can change the prefitler in centWave. I don't have much experience with the online version, so I don't know if there are other added parameter that could help. To set the prefilter in the online version I would change the setting using the edit tab button. Then under... feature detection there's an advance link. After clicking this you get a option for changing the pre filter. The pre filter is two numbers the first is, as they call it pre filter peaks, is the minimum number of scans the peak has to be seen in. So if your peak is very short and your scan time is long, then maybe you need to set this to 2. The second number is, as they call it pre filter intensity, is the minimum counts for the peak. I would check that the peak is over 100 if not set it lower. There is also a signal to noise threshold, which you might want to lower.
It also seems that there is some sort of cutoff for the peak width so that you can't put the value below 5sec. I hope that all of you peaks are longer than 5 seconds. If the peak isn't then this is why you're not seeing the peak. It seems that there isn't an option for using the other grouping algorithm 'nearest'. I would check the npeaks column at the end of the run to check that two or more features haven't been grouped together.
Let me know how you get on and sorry again for the confusion.
Paul
Hi Paul,
Thanks so much for your reply, and your answer definitely helps.
My parameters are already set so that I can capture as much stuff as possible. When I use Marker Lynx, my ion comes through. So if I were to change signal/noise threshold (which is currently set at 10) what do you suggest I put it at?
Thank you very much!
Majda
majah,
I would try s/n=5. What does the peak look like in marker lynx? Does the peak look normal, i.e. gaussian?
Let me know how you get on.
Paul
Hi Paul,
I played with parameters and set s/n=5, but still nothing, I cannot get the mass to come through in XCMS. The peak looks normal, yes!
That particular mass has relatively abundant counts in 30% of samples. So it really should be coming through.
I really don't know what other parameters to tweak so that I can get it to come through? Any suggestions?
Thank you so much for your help!
Majda
bw is pretty high for UPLC data.
Try using the regular UPLC/QTOF parameterset and only lower minfrac if you know that the feature you are looking for is present in less than <50% of the samples.
Thank you very much! I will try that, and hopefully I can get it!
Hi Ralf,
I decreased both the minfrac and bw and I still cannot get that particular mass to come out. Do you have any additional suggestions?
Thank you very much!
Majda
The only thing I can figure is that it is overly saturated and therefore the accuracy is very poor.
If you want you can send me one of those files where the peak should show up and I will have a look at it.
Hi Ralf,
Thank you for your help. I emailed you a file to look at!
Majda
Hi Ralf,
Did you receive my email with a file attached?
Thank you!
Majda
Yes, I will have a look at it.
The intensity of this feature is extremely low (~200) and it consists of only 3 data points above the baseline (centWave requires a absolute minimum of 4 data points)
It looks much more like a spike than a feature and is therefore rejected by centWave.
It can be detected by reducing the minPtsAboveBaseLine parameter to 3, but this parameter is currently not changeable in XCMS or XCMS Online.
Hi Ralf,
Thank you for looking into this and for your reply. Is this parameter "minPtsAboveBaseLine" going to become changeable in the near future?
Thank you very much!
Majda
Yes, it will probably be added as an extended option.
Hi,
it has been a while since I posted. I just have a quick question regarding the "minPtsAboveBaseLine" option, and if it is available for tweaking in the online XCMS version at this time? This is important for some features I would like to be able to extract using XCMS.
Thank you!
Majda
Thank you for the reminder. We'll add that option soon.