MSstats tutorial from the May Institute, 2020
Based on the following video; https://www.youtube.com/watch?v=30a6sDj8Qlo&list=PL2u38g_AG4MG_puKZBT4HawiVLQBE28gD&index=10
Github for tutorial; https://meenachoi.github.io/MayInstitute2019RstatsMS/
Data from the following; https://docs.google.com/document/d/1vnMG0XTqYVngpK5EyIQNXi-XDDh7sq72yQnhrnxw83k/edit
Further talks found here; https://computationalproteomics2020.khoury.northeastern.edu/
And talks from 2022; https://computationalproteomics.khoury.northeastern.edu/
Google group for troubleshooting MSstats;
https://groups.google.com/g/msstats
It might be easiest to do basic QC using the artMS package; https://www.bioconductor.org/packages/devel/bioc/vignettes/artMS/inst/doc/artMS_vignette.html#1_OVERVIEW
-
Following the comparison testing, there are some issues. Some proteins have p-values and adjusted -pavlues of exactly 0, with infinite fold change. I have removed these, however I'm uncertain if this is the appropriate thing to do. I should discuss this with Lisa Woods. I've since seen this on the Google groups for MSstats, which say to replace p-values of 0 with a very low constant. This is essentially due to floating-point error for very low numbers. See more here: https://groups.google.com/g/msstats/c/50cmwFpK-ho/m/V_BMSZyF3MYJ There are also some proteins which give a p-value/adjusted p-value of NA, and I've also removed these.
-
There is a series of warnings in the data-import and tidying step; "Warning in aggregator(Intensity, na.rm = TRUE) : no non-missing arguments to max; returning -Inf", which I should investigate further.
-
I would also like to add protein names in the format of gene names to replace the uniprot accession numbers.
-
Funny volcano plots noted, see discussions here:
https://www.researchgate.net/post/Artefact_in_volcano_plot
https://groups.google.com/g/msstats/c/siW4LZc0KLk
try running MSstatsGroupComparison instead of groupComparison in the comparison section.
- Currently I'm struggling to get the function
PDtoMSstatsFormatto work. Initially I was missing the columns for "ProteinGroupAccessions" and "PrecursorArea", however I renamed the column"Master.Protein.Accessions" from the original dataset. "PrecursorArea" can be renamed from the "Precursor.Area" or "Precursor.Abundance" column, but note that these are only produced when running a quantified search in Proteome Discoverer.
I've solved these problems by running my samples simultaneously in Proteome Discoverer as a lable-free quantification experiment, only then does it include information on precursor abundance. The other missing columns can be created by renaming existing columns.
- However, I'm still having other issues, being discussed here: https://groups.google.com/g/msstats/c/fHa3MqRtMss/m/PCLGjy0SCAAJ
These problems solved, I'm not sure how, but likely by updating package dependancies.
- I'm having trouble running the comparison matrix, and am only able to do the "pairwise" analysis. This may well be appropriate for my study however. This I solved by reading the
MSstatstutorial, the contrast matrix is supposed to have columns not for every sample, but for every condition. These also need to be listed in alphabetical order.
Feel free to use this code as you wish under the MIT license, however an anknowledgement would be nice. Thanks!