Hi everyone,
I am currently undertaking a differential expression analysis over pseudotime on embryonic scRNA-seq data.
I used tradeSeq to identify genes that are differentially expressed across pseudotime in my lineage. Now I was wondering, are there existing tools to identify Gene Ontologies that are over-represented over pseudotime, considering that each gene has its own time frame of expression peak ?
For instance, I created this heatmap where we can identify the specific timing of expression of each DEG over pseudotime. I believe performing gene ontology ORA on genes that are highly expressed during the earliest pseudotimes would be very informative, but it would be less clear thereafter.
Would it be a good idea to define specific timeframe of expression, considering the phenotypes of cells (embryonic stage, etc...) or would it be a biased approach to do this ? I am eager to have your point of view on this subject.
Can you clarify your question of interest a bit? I.e what exactly are you trying to understand/see/show with the GO term overrepresentation analysis?
Sure thank you for your reply. What would be interesting to me is to see the sequential transcriptome switches of GO pathways. The idea is to summarize important GO pathways for early embryonic development and show their sequential enrichment. Ideally I could summarize this information with a plot like this (from Lv et al. 2019 Plos Biology) :
I believe the approach that would fit the most to what I would like is to cluster genes according to their expression pattern across pseudotime and perform gene ontology on it.