how to use DESSO to my own data? about desso HOT 4 CLOSED

Hi Yaoyong,

Based on your email and this github issue, I would like to clarify more details as follows:
(1) The train.py is used to train a CNN model based on the given ChIP-seq data. After training, we will get motif detectors which are actually the convolutional filters in the trained model. This original code is designed for the 690 ChIP-seq datasets as we mentioned in the paper. We also have the code designed for user's own ChIP-seq data but need to be tested before uploading to this repository. Another co-author is working on this and will let you know once ready.

(2) We have the trained models for the 690 ChIP-seq datasets. However, each model is trained specifically for each dataset, so it may not very make sense to identify motifs from your own data by using predict.py. My suggestion is to train a new CNN model for a new dataset then make predictions.

Feel free to let me know if you might have any other concerns. Thanks.

from desso.

Dear DESSO Developers,

thank you very much for the rapid and comprehensive responses to my question. I am looking forward to the codes which I can use to apply DESSO to our own ChIP-seq data.

Best regards,

Yaoyong

from desso.

Dear DESSO Developers,

thank you very much for the rapid and comprehensive responses to my question. I am looking forward to the codes which I can use to apply DESSO to our own ChIP-seq data.

Best regards,

Yaoyong

@Wang-Cankun Cankun, have you finished the code test?

from desso.

Hi @yaoyong02

@Wang-Cankun finishes the code test and provides the tutorial for how to apply DESSO to your own ChIP-seq data. Please let as know if you might need additional help. Thanks.

from desso.