Hi, I'm doing a principal component analysis for an RNA-seq dataset in Seqmonk and trying to determine which data points (genes) contribute most to each axis/component. I understand that the genes with the highest and lowest transformations (rotations) are the ones contributing most strongly to each axis, but I was wondering what, if anything, these highest and lowest transformations represent. Taking this into account, when trying to determine my "most highly contributing" genes, should I be choosing (a) my top x genes from either 'end' of the transformations axis, or (b) genes from either end of the axisabove a certain absolute value on the transformation axis, irrespective of whether there are more on one 'end' than the other? Additionally, would you recommend a particular 'threshold value,' or is it entirely up to the aim of the experimenter? Thank you!
