osu-bmbl / desso Goto Github PK

Dear DESSO Developers,

my name is Yaoyong Li. I have been reading your paper about DESSO and have also downloaded the software. Your method looks promising. I would like to apply it to our ChIP-seq data. However, I am not sure how I can apply the software to my own data, because I could see any option for the user's own data as some input in either train.py or predict.py, two main python scripts in DESSO. In another word, I suppose that the input should be a list of DNA sequences obtained from the peaks of my ChIP-seq data, and I wonder how I could give the list of DNA sequence to the software.

Best regards,

Yaoyong

Website is not accessible anymore:

https://bmbl.bmi.osumc.edu/DESSO/index.php
https://bmbl.bmi.osumc.edu/DESSO/
https://bmbl.bmi.osumc.edu/desso/

Hello,

I want to predict the TF binding sites using your example peak file. The reference genome is hg38, and the bed file is fox01_peaks.bed.
After processing peak file using command "python processing_peaks.py --name fox01_peaks.bed --peak_flank 50", I started to train the model. but i got an error message below.

python train.py --start_index 0 --end_index 1 --peak_flank 50 --network CNN --feature_format Seq Sequence length: 101 Dataset: fox01_peaks_encode Traceback (most recent call last): File "train.py", line 163, in <module> tf.app.run() File "/home/zhenyingLab/zhukangli/.conda/envs/desso/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 48, in run _sys.exit(main(_sys.argv[:1] + flags_passthrough)) File "train.py", line 68, in main util.load_data_encode(PATH_ENCODE + "/" + train_data_name + "_AC.seq.gz", peak_coor, train_data_name, path_curr_data, "train", DNA_SHAPE, peak_flank, back_grou) File "/storage/zhenyingLab/zhukangli/software/DESSO-master/data/libs/util.py", line 138, in load_data_encode sequences = np.concatenate((sequences, seq_shuffle), axis = 0) ValueError: all the input array dimensions except for the concatenation axis must match exactly
Any help or guidance would be greatly appreciated!

Thank you!

kangli zhu