Common Cutoff in TPM unit to filter out gene considered as low expressed.

Question

[ZheFrench]

Question is in the title.

For example you computed "estimated expression" in TPM units per gene for one biological replicate. (with Salmon for example).

In general, what can you admit as a cutoff to remove low expressed genes.

I think I saw 1 was an admited cutoff but not sure. For now I'm filtering by 50th percentile...

Answer 1

[Devon Ryan]

There's no single cut-off threshold that can be used. You'll need to use a method like zFPKM to determine what's appropriate in your dataset.