1 Topic by Julian

  • Julian
  • Member
  • Offline

Topic: Averaging of multiple spectra

Hi there,

first things first, thank you for all the work you're putting into mmass!
Its a great piece of software (as you know)

I've run into a problem though.
I've got some peptide mass fingerprinting spectra into my hands.
They were recorded  using a Thermo Fischer Xcalibur Orbitrap, deconvuluted and converted to msxml.
1 measurement equals to roughly 200 spectra.
I'm able to load all of them -- as soon as I try to average them mmass crashes.
No output on cli or in the log files is provided.

I am able to average them ten at a time and create a "total" by averaging the averages.

Has anybody run into this problem before and found a solution?

2 Reply by silenj

  • silenj
  • Member
  • Offline

Re: Averaging of multiple spectra

Hello,

I have been using the fantastic software and making modifications of it to include COSIMA/Rosetta spectra into it. Reading spectra worked out to be doable, but I run into a similar kind of problems with the Math tools. Either the system crashes or the results are incorrect. There is some strange effect for low masses < about 80 Da which point in a direction of a +1 offset error somewhere.

When the program crashes the --sync option gives a message indicating problems in the fast c-code codeset which I have not yet been able to understand.

As a summary, I have had exactly similar problems. The errors seem to be real.

Regards Johan

3 Reply by Martin

  • Martin
  • Administrator
  • Offline

Re: Averaging of multiple spectra

Hi,
First of all I have to mention that mMass was never designed to handle huge amount of spectra such as whole LCMS run at once. It is more dedicated to pick individual spectra of interest for detailed manual analysis. With respect to that I am impressed that you are even able to open 200 spectra :-)

It is not clear from your post whether you are working with profile or centroided data. Both have their own limits. For the profile data the problem is typically a different raster, so each time you join two spectra you may get the number of points which is the sum of both spectra. If you do this 200 times with high-resolution data… it probably crashes because of memory error. Similarly, for centroids it can be problem to display resulting mass list in the UI. Especially on a Mac this is a known issue.

The mass shift you have observed is a different topic. This is very strange. If you can share some example data to reproduce this issue I would be happy to check it.
martin

4 Reply by silenj

  • silenj
  • Member
  • Offline

Re: Averaging of multiple spectra

[quote=Martin]Hi,
First of all I have to mention that mMass was never designed to handle huge amount of spectra such as whole LCMS run at once. It is more dedicated to pick individual spectra of interest for detailed manual analysis. With respect to that I am impressed that you are even able to open 200 spectra :-)

It is not clear from your post whether you are working with profile or centroided data. Both have their own limits. For the profile data the problem is typically a different raster, so each time you join two spectra you may get the number of points which is the sum of both spectra. If you do this 200 times with high-resolution data… it probably crashes because of memory error. Similarly, for centroids it can be problem to display resulting mass list in the UI. Especially on a Mac this is a known issue.

The mass shift you have observed is a different topic. This is very strange. If you can share some example data to reproduce this issue I would be happy to check it.
martin[/quote]

Yes, this is fine and I think this is the correct explanation. The missing low masses seem to be due to some feature of the crop function. If crop is selected, the feature disappears. For my application this is sufficient. I am able to sum about 10 spectra which is about 1G of numbers.

Regards Johan

5 Reply by Julian

  • Julian
  • Member
  • Offline

Re: Averaging of multiple spectra

Hey,

thanks for the answers! I'm working with profile data, so this may explain the crash when averaging the spectra.

I've worked around the problem by averaging the spectra directly after measurement, so my problem is closed ^^

Thanks again,

Julian