How to filter out an RNA-seq differential expression data set?

Question

[pvaysberg]

Hello, I have a data set with expression values (counts). I assembled the transcriptome myself, since there is no reference to map to. I started with 130,000 transcripts (rows on my data table) and brought it down to about 70,000 by filtering out all transcripts with (sum of row FPKM) < 1. This was to eliminate lowly expressed gene and stochastic noise. I don't know how to further filter my dataset to only find differentially expressed genes, unfortunately. Any tips or resources?

Answer 1

[Devon Ryan]

You need to calculate adjusted p-values and filter according to that. With FPKM, people often use an ANOVA or similar linear model.