PCA from kallisto output

Question

[Picasa]

Hi,

My goal is to plot a simple PCA to compare samples after kallisto run.

I am a bit confuse how to do it. Should I take the "est_counts" or the "TPM" values ?

[biofalconch]

Just taking into consideration what the package rnaseqGene has to say, I would use normalized data to to a PCA. Hope it helps :)

[Picasa]

From what I understand, TPM is already normalized no ?

The package you mentioned seems to recommend to apply rlog on counts data.

Answer 1

[informatics bot]

My lab takes the estimated counts, rounds them to the nearest integer, performs VST normalization using the DESeq2 package, then plots the first two PC coordinates from the prcomp() function.

[poisonAlien]

Better to use tximport instead of manually rounding off.