Entering edit mode
marak
•
0
@marak-18210
Last seen 5.9 years ago
i am new to RNA seq analysis .I have TMM normalized htseq-count data .Can any body tell can me what threshold should i use inorder to differentiate between genes related to cancer and that are normal?i.e i want to know which genes are differently expressed using machine learning algorithms(PSO+SVM)
Machine learning algorithms aren't designed for differential expression, but instead are for sample classification. Is there any particular reason why you think you should be using SVM for that sort of thing?
ok i want to use it for identification of cancer related genes
Hi Marak, fellow newbie here. I think differential expression workflow is enough if you only want to find cancer-related genes by comparing cancer against normal tissues (just an example). Simple methods are fine too, you know.
Before you're diving in too deep with all those fancy machine learning algorithms out there, keep in mind that these techniques work best if you have large enough samples or you've already prepared to validate your result in an independent cohort.