I am going to write this as if you are able to see named compounds, Ion ratios, some integration and qualifiers. I am going to explain how to analyse the information between specs. I am not so savvy in applied statistics so that is where Ill stop and hopefully someone else will jump in.

If you are able to see the integration patterns from each spec, I would make first to check the Ions per compounds that you are looking for. There is no way to know how the person in the study integrated each peak you just assume it is represented honestly and correctly. You also need to make sure that the reported data is within the same units, Make sure it is not a collective of a mean or average when analyzing numbers. This can throw information off. If it is see if they reported raw data. The worst and simplest mistake to do is to get a chromatography sample and add it up all wrong. If your are able check the ions present in the reported compound and check ion ratios you can be sure that the evaluated compounds are the same that you are comparing. Also some ions can seem way off and depending on how low they calibrate some ions have horrible recovery so make sure of there is a wide variation between your spec. look to check any qualifiers your studies may have mentioned. Some very slightly but even stock information can be beneficial in depicting the ions and patterns. Although you will not know their calibration to know where their control values came out at on their spec, the reported information should be consistent. They should provide the internal standards that they used and the limits +/- 15% for IS counts is pretty standard depending on the compounds you are looking for. This check should make it easier to compare the different specs from different research, all-compounds should share the same ions and ion ratios. Proceed from there. That is how I have compared many spec analysis. It takes some time but it is a good way to insure you are collecting the most honest data.

If you have the raw spec and you are the one looking over each study, I would get as much information as you can. Ask for their calibration and LOQ numbers. Ask for a printout of their full qc for that run and ask for a copy if the CCV and CCB, this is a good way to check the most accurate information. Check retention ,ingratiation, ask for any qualifiers and ask if there are any compounds that are reported that have a history of shifting overlapping or bad recovery to make sure nothing is doubled up.

This should make the studies and even playing field and easier to evaluate and collect your data accost the board.

I hope I was able to answer your question on how to compare data from individual studies accosts different spec.

I think you are looking for a Multivariate Analysis software package: https://www.statisticshowto.datasciencecentral.com/probability-and-statistics/multivariate-analysis/ This link will give you a description - if this sounds like the right place to start you can search R and/or Python for publicly available packages to perform the statistics.

I would read over the reports that the ABRF PRGs have done where they designed studies specifically to compare qualitative and quantitative proteomics data across equipment, labs and protocols -- often with specially designed samples. It wont answer all of your questions but will provide background to a lot of the considerations especially in terms of MS quantitation strategies.

https://abrf.org/research-group/proteomics-research-group-prg

The latest work from CPTAC https://proteomics.cancer.gov/programs/cptac also has some good pointers https://proteomics.cancer.gov/sites/default/files/assay-characterization-guidance-document.pdf

Thank you Jennifer Papuga Kerri Ball and David Bramwell

The suggestions were a lot of help and will definitely help me in future which cleared my knowledge about the additional facts provided.

More specifically in terms of the statistical analysis I would look at

https://cran.r-project.org/web/packages/rmeta/rmeta.pdf

and

https://cran.r-project.org/web/packages/forestplot/vignettes/forestplot.html

and as a parting note I would say that currently, in my opinion, analyte quantitation in MS (on complex proteomics samples) is not comparable across labs (unless they expressly had it as a goal of a study as in the CPTAC case) sometimes it is not comparable across different MS platforms within the same lab.

In terms of meta analysis you may be better working at the 'conclusion' level i.e. these analytes were 'significantly' differentially expressed. In some cases you may be able to compare ranks of analytes across studies.

David Bramwell Thank you so much! It was much of a help and quite agree with your view.

Hello Varshasnata,

You asked, and some one listened to just you!

There you go, a new Meta Data analysis R package for "proteomics/mass-spec data"! : )

MetaMSD: meta analysis for massspectrometry data : https://peerj.com/articles/6699.pdf

Could not get any better!

Thanks,