search

BackofenLab / Cherri

https://backofenlab.github.io/Cherri/
GNU General Public License v3.0
0 stars 0 forks source link

Define User interface #35

Closed teresa-m closed 1 year ago

teresa-m commented 2 years ago

Generate wrapper, which can handle as input sequences or a given interaction. Generates features for it and then calls a model. In the end, we would have a model file, an input- and used-features file, and the wrapper script.

teresa-m commented 2 years ago

The script should have two start points:

  1. from Chira output to generate a training set! -> Wrapper create_trainings_data.py
  2. only generate positive instances -> Wrapper evaluate_Instance
teresa-m commented 2 years ago

df what colums do we need:

chrom_1st start_1st end_1st strand_1st

interaction_no

chrom_2end start_2end end_2end strand_2end

'score_seq_1st_side', 'score_seq_2end_side',

'biotype_region_1st', 'biotype_region_2end',

teresa-m commented 2 years ago

test both wrapper with a bigger dataset:

  1. python create_trainings_data.py -i1 /vol/scratch/data/RRIs/Paris/  -g /vol/scratch/data/genomes/hg38_UCSC_20210318.2bit -r ChiRA_interaction_HEK293T_1.tabular ChiRA_interaction_HEK293T_2.tabular ChiRA_interaction_HEK293T_3.tabular  -o /vol/scratch/data/test/ -c 100 -n test_paris_human -l /vol/scratch/data/genomes/hg38_Info.tab
  2. python evaluate_instance.py -i1 /vol/scratch/data/RRIs/SPLASH_without_hybrids/ChiRA_interaction_summary_ hES_RA_2.tabular -i2 none -g /vol/scratch/data/genomes/hg38_UCSC_20210318.2bit  -o /vol/scratch/data/test/ -c 150 -n test_data -l /vol/scratch/data/genomes/hg38_Info.tab