'combined score' calculation with singscore vs gsva
1
0
Entering edit mode
rocanja ▴ 60
@rocanja-5374
Last seen 8 days ago
Australia

For 'directional' gene sets (eg. where we have a gene set that contains the UP regulated genes after a particular stimulus and a corresponding gene set with the DOWN regulated genes), the singscore package has an implementation that calculates a 'combined score' (TotalScore) from the two individual scores that can be interpreted as the 'concordance' with a particular 'phenotype'. Mathematically this is achieved by summing of the individual scores (conceptually, I think it's done as A+(-1*B)). I was wondering whether something similar can be applied to gsva scores to obtain a 'combined gsva score' for a pair of directional gene sets? Since the algorithms behind the gsva and sing scoring are different, I would like to get some feedback on whether a 'combined gsva score' could conceptually be interpreted similar to the 'combined singscore'?

singscore GSVA • 1.4k views
ADD COMMENT
1
Entering edit mode
Robert Castelo ★ 3.1k
@rcastelo
Last seen 1 day ago
Barcelona/Universitat Pompeu Fabra

Regarding the scores calculated with the GSVA package using the default method (method="gsva"), because a positive enrichment score indicates that the expression of the genes forming the gene set have higher expression ranks than those forming a gene set with a negative enrichment score, I guess that you could combine the scores of gene sets in the way you describe to verify what you call the "concordance" of a phenotype with the description of those two gene sets. Single-sample gene set enrichment scores are more a starting point of an analysis than the end, so you can envision multiple ways in which those scores can be harnessed to learn about the biology of the samples you are analyzing. If you feel the GSVA package should include a new functionality, please open a new issue in its GitHub repo describing with as much detail as possible what would be that feature request.

Best regards,

robert.

ADD COMMENT
0
Entering edit mode

Thanks so much @rcastelo for your feedback - that sounds promising, I'll give it a go ... and it would be fantastic to have the 'combined gsva score' implemented into the GSVA package! I have been using gsva scores for many years to assess gene set expression and changes in our RNAseq data sets. And I am also using the singscore package, especially for the feature of the 'combined score', as it makes visualisation more efficient ... instead of plotting 4 scores from 2 directional gene set pairs pairwise against each other, it can be summarised into one scatter plot showing the two combined scores. But it only occured to me now that I could try to derive a 'combined score' from the gsva scores as well!

ADD REPLY
0
Entering edit mode

Ok, I've added a new feature request issue about this to the GSVA GitHub repo. If you are (or want to become) a power user of GSVA, and of any other Bioconductor package or software in general, I strongly recommend you to create your own profile in GitHub to be able to open issues by yourself on the GitHub repositories of open-source software, such as Bioconductor packages, which helps developers to keep track of what's pending and, very importantly, GitHub issues also help engaging users in conversations about new features and bugs, which are crucial to meet users expectations and needs on the software. I'd appreciate if you can provide details on your own use case of this feature and what would you like to get from GSVA through the issue I've open on this feature request.

ADD REPLY
0
Entering edit mode

Thank you so much! I am on GitHub, just hadn't gotten around to log the feature request yet!

ADD REPLY
0
Entering edit mode

Just in case this wasn't clear, I've already added the feature request, you just have to log into GitHub with you account, go to this link and leave your comments about how would you like this new feature to be, if possible with a motivating example from some publication. Thanks!

ADD REPLY

Login before adding your answer.

Traffic: 629 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6