search

kundajelab / chrombpnet

Bias factorized, base-resolution deep learning models of chromatin accessibility (chromBPNet)
https://github.com/kundajelab/chrombpnet/wiki
MIT License
124 stars 34 forks source link

Does ChromBPNet normalize the input data at any point? #204

Closed idf-io closed 3 months ago

idf-io commented 3 months ago

Dear ChromBPNet team,

I'm using ChromBPNet for my project and it's essential for me to know if ChromBPNet normalizes the ATAC-seq data as part of the workflow.

Currently I am using your reads_to_bigwig.py tool from the preprocessing section to create the bigwigs from my ATAC-seq fragment reads.

I couldn't find any normalization procedure within the pre-processing code. Does ChromBPNet normalize during training?

I would be very thankful for your help.

Thank you in advance and best regards, Ian

akundaje commented 3 months ago

No it is trained on raw counts and predicts expected counts per base. You can normalize as you wish post hoc

On Tue, Aug 20, 2024, 5:14 AM Ian Dirk Fichtner @.***> wrote:

Dear ChromBPNet team,

I'm using ChromBPNet for my project and it's essential for me to know if ChromBPNet normalizes the ATAC-seq data as part of the workflow.

Currently I am using your reads_to_bigwig.py tool from the preprocessing section to create the bigwigs from my ATAC-seq fragment reads.

I couldn't find any normalization procedure within the pre-processing code. Does ChromBPNet normalize during training?

I would be very thankful for your help.

Thank you in advance and best regards, Ian

— Reply to this email directly, view it on GitHub https://github.com/kundajelab/chrombpnet/issues/204, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABDWEKEB2NWDC53YFIG4Z3ZSMXJHAVCNFSM6AAAAABMZYSUUWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGQ3TKNBZGM4DQNY . You are receiving this because you are subscribed to this thread.Message ID: @.***>

idf-io commented 3 months ago

Thank you!