How are high absolute attribution scores defined?

jmschrei / tfmodisco-lite

A lite implementation of tfmodisco, a motif discovery algorithm for genomics experiments.

MIT License

56 stars 16 forks source link

How are high absolute attribution scores defined? #47

Open FelixWaern opened 8 months ago

FelixWaern commented 8 months ago

When extracting seqlets with high absolute contribution scores, are the seqlets considered to be high in comparison to all sequences contribution scores or only within it's own sequence?

jmschrei commented 8 months ago

High w.r.t. all sequence contribution scores. A Laplacian null is calculated over all scores and used to FDR correct seqlets. https://github.com/jmschrei/tfmodisco-lite/blob/main/modiscolite/extract_seqlets.py#L17