TFModisco

Here we discuss the TFModisco command, which conducts de novo motif discovery

This function integrates the tfmodisco-lite motifs and tfmodisco-lite report commands, that leverage the TF-MoDISco algorithm to identify the top sequences with high contribution within a defined set of peaks. For a more detailed explanation of the function use, please refer to the tfmodisco repository.

Usage

The following function will only run if ContribsBW has been run previously, as the TF-MoDISco algorithm pulls from the contribution scores generated.

TFModisco \
   "$output_directory" \
   "$jaspar_file" \
   "$condition" \
   "$n_seqlets"

Input format

output_directory    # Cohort directory including the cell types with pre-computed contribution scores (e.g., ~/ChromBPNet/outputs/mouse)
jaspar_file         # Input database of motifs (e.g., JASPAR_CORE_2026_non-redundant.meme) - must be in MEME format
condition           # Name of the condition/parameters tested (e.g., if multiple runs were conducted using different n_seqlets)
n_seqlets           # Number of seqlets used in the analysis (e.g., 100000) - See Note

Note: For all single-cell ATAC-seq models, 1M seqlets was used as input. For the bulk ATAC-seq models, only 100k was used, as it sped up the processing and were used for validation purposes.

Output format

All output files will be found in ~/ChromBPNet/outputs/$cohort/$cell_type/tf_modisco/

modisco_report_{condition}\      
	*.png                     # Images of high contribution motifs identified and associated with a known database motif
        motifs.html               # HTML document with all identified high contribution seqlets and 
        trimmed_logos\            # Images of trimmed high contribution sequences identified 
modisco_results_{condition}.h5    # File including all seqlet information identified by the TF-MoDISco algorithm

Downstream analyses of the TFModisco results and explanation of the file contents can be found in Downstream_analyses/motif_analysis.