search

Nesvilab / FragPipe

A cross-platform Graphical User Interface (GUI) for running MSFragger and Philosopher - powered pipeline for comprehensive analysis of shotgun proteomics data
http://fragpipe.nesvilab.org
Other
184 stars 37 forks source link

Label-free quantification #80

Closed marcisaksson closed 5 years ago

marcisaksson commented 5 years ago

Describe the problem

I have run several raw-files in fragPipe with an open search, and I enabled label-free quantification. According to my understanding, spectral counts are used for this quantification right?

The problem is, I don't know where I can extract the spectral counts from different files (samples), for the relative quantification? The report.tsv-file seems to report the total number of spectral counts from all runs which is not very useful for me. I would like to separate the samples/runs into two groups, and run a t-test to find differentially expressed proteins using the spectral counts.

Last thing, how to best deal with the normalization of the spectral counts prior to running a t-test?

Best,

Marc

System info

You can find that printed on the Config tab.

Describe your experiment

Genral proteomics experiment description

e.g. "TMT, Human, full cell lysate with Trypsin" , "AP-MS pulldowns, mouse, liver tissue"

...

Input data files

e.g. "fractionated HeLa, 5 samples, 3 bio replicates, 2 technical replicates" or "ten 3 hour LC gradients, full cell lysate, fruit fly" or or at least "5 mzml files 1.5Gb each"

Sequence database

your response here, preferably a link or at least name, organism etc.
Size of the either in proteins or in megabytes

Attach fragger.params file

You can find it in the output directory you specified for analysis.

Run log output

** copy text from the text console in the gui here **
prvst commented 5 years ago

@marcisaksson

Unfortunately fragpipe does not provide a way to analyze your files in a separate way, I guess you would have to run the program multiple times, one for each sample or group of files. If you are OK with running programs on the command line, you might want to give a try to Philosopher, it allows you to separate your spectra files into groups or samples.

marcisaksson commented 5 years ago

I understand. I guess it could be put on a 'to do list' for fragPipe, since it integrates msFragger and Philosopher already. It would be neat to have the option to set an experimental design as in MaxQuant, resulting in an output with quantitative intensities for each samples (similar to LFQ intensities in MaxQuant). In the mean time, I will look into running philosopher with cmd, even if I am an absolute beginner.

chhh commented 5 years ago

@marcisaksson So the feature you want is to be able to group sets of input LCMS files?

chhh commented 5 years ago

@marcisaksson Also, to simplify your start with running everything from the command line, in FragPipe check the Dry run checkbox next to the Run button, and click Run. It will just print the commands to the text console, so you can copy paste them into your scripts as a starting point.

chhh commented 5 years ago

This is related to #31

marcisaksson commented 5 years ago

Actually, it's more like being able to add the files as separate samples in fragPipe. The report.tsv file could then report the sample specific spectral counts even if it might mean many extra columns. This is similar to what happens in MaxQuant with LFQ intensity columns, when setting different sample/experiment names to the raw files.

chhh commented 5 years ago

@marcisaksson So it's a little more. You want to be able to label single files - call it sets, experiments, samples - it's some way of grouping input LCMS files. Then produce a combined report which includes columns for each group. @prvst Does philosopher support that? Sounds like a good feature to have - comparative analysis is very common.

marcisaksson commented 5 years ago

Yes, we could call it grouping. I will continue with the MaxQuant analogy here :) In MaxQuant, you are able to set individual experiment names to individual raw-files. That will produce LFQ intensity columns for each individual run.

Alternatively, you can set the same experiment name to several raw-files (in case of prefractionated injections from a single sample, or technical replicates). In the latter case, MaxQuant will sum the intensity for these runs and produce one LFQ intensity column (since the experiment name is the same).

That is a feature I like, but I don´t know how difficult it is to implement, especially if there should be normalizing steps prior to summing the counts for the raw files having the same experiment name.

chhh commented 5 years ago

Fixed in v8.4 (not yet publicly released)