I am using Hisat2 and Stringtie for alignment and assembly of human samples. When I run ballgown on my data, I am getting 29 genes that are significantly differentially expressed, however when I use DESeq2 for the analysis, I get 930 genes that are significantly different (q<0.05). I also compared the top 100 genes sorted by q value for both ballgown and DESeq2, and only 17 genes are the same between the two.
My question is, why am I getting so few genes showing as significantly different with ballgown? If it were a problem of multiple sampling, shouldn't I have more overlap between DESeq2 and ballgown when they are sorted by q-value?
In ballgown, I am using a coverage cutoff of 10 to filter my data before running the stat test. I've tried other values and methods for filtering my data in ballgown, and this is giving me the best result so far.
Thanks for your help!