An RBP motif prediction algorithm based on target site accessibility

This motif discovery algorithm selects the final motif with most distinguished feature (total accessibility or site number) in bound (positive) and unbound (negative) transcripts.

Pre-requirements

Python 2.7
RNAplfold (version 2.1.8)

Sequence accessibility of transcripts is predicted by the RNAplfold tool in the ViennaRNA package.

Input files

pos_file: A file that contains gene names in the positive set, one gene per line.
neg_file: A file that contains gene names in the negative set, one gene per line.
seq_file: A fasta file containing sequences of transcripts in positive and negative sets (in the RNA alphabet). Make sure gene names in this file are consistent with those in pos_file and neg_file.
RNAplfold_direct: A folder containing the <ID>_lunp files of RNAplfold output of genes in positive and negative sets. Make sure gene names in this file are consistent with those in pos_file and neg_file.

To get result from RNAplfold, assuming the length of the binding site is at most 10 nt:
```
RNAplfold -W 80 -L 40 -u 10  < seq_file_name
```
The setting of parameters W, L, u can be changed upon different situation.

For detailed information of running RNAplfold, please refer to https://www.tbi.univie.ac.at/RNA/RNAplfold.1.html.

Parameters

To run the code, you will need to specify the following parameters by modifying the following lines of run2_new_RNAplfold_format.py:

model = 'access_seq'
pos_file_name = ''   # name of pos_file
neg_file_name = ''   # name of neg_file 
seq_file_name = ''   # name of seq_file
RNAplfold_direct = ''  # directory name of the RNAplfold results

final_out_name = ''   # name of the final output (only the best motif)
detailed_final_out_name = ''   #name of the detailed output (motifs from all five seed 6mers)

Motif prediction

To run the code, one need to run the following function in the run2_new_RNAplfold_format.py,

Motif_discovery(final_out_name, detailed_final_out_name, model, pos_file_name, neg_file_name, seq_file_name, RNAplfold_direct)

Related Publication

X. Li, G. Quon, H.D. Lipshitz, Q.D. Morris, Predicting in vivo binding sites of RNA-binding proteins using mRNA secondary structure, RNA 16.6 (2010): 1096–1107. [Pubmed] [Supplementary materials]

morrislab / ATS-motif-prediction