This repository contains code and related datasets for a new method to make RNA-protein partner predictions.
The "Data" folder contains: (i) Interacting_RNA_Protein_Pairs.txt - this file contains PDB IDs and the IDs of protein and RNA chains found to be interacting using the 5 angstrom distance cut-off.
(ii) NPInterUniqSeq.txt - 11,281 positive examples of known RNA-protein pairs. The file is in the following format:
sequence_name Protein sequence RNA sequence
(iii) NRBP.txt- 971 negative examples i.e. proteins which are not known to bind RNA, randomly paired with RNAs from fRNAdb. The file is in the following format:
sequence_name Protein sequence RNA sequence
(iv) Positive_Fasta_RPI2241.txt - positive examples
(v) Negative_Fasta_RPI2241.txt - negative examples
Files (ii) and (iii) constitute the independent test set. Files (iv) and (v) constitute the training set.
To run RPIMotif-ct, you will require the following files, as well as Perl: (i) less_motifs.txt (ii) arffHeaderP3R4 (iii) RF_17910_ct.model (iv) createARFF_test.pl (v) runModelonTest.pl
You first need to create a test arff file with your query RNA-protein pairs. The format in which the data should be provided is shown in test_data.txt. Run the following command to generate an arff file: perl createARFF_test.pl less_motifs.txt test_data.txt
The output from this will be test_data.arff
Next, run the following command: perl runModelonTest.pl RF_17910_ct.model test_data.arff
The output will be in Output_step2.txt in the format: Query_Number,Probability,?,Prediction
The prediction will be either "0" which means the RNA-protein pair is not predicted to interact or "1" which means the RNA-protein pair is predicted to interact.