Software
Source codes are provided to run beRBP in the local Linux computer. A job can be submitted to the beRBP webserver if less than 5000 candidate sequences.
- beRBP.tgz
- beRBP.standalone.README.txt
- hg38.phyloP100way.bw This file (of size 10G) contains conservation scores for human genome hg38. Download it here or from http://hgdownload.cse.ucsc.edu/goldenpath/hg38/phyloP100way/hg38.phyloP100way.bw, and put it at the uncompressed directory beRBP/lib/hg38.phyloP100way.bw.
Genome-wide prediction results
We have generated genome-wide prediction results for 26 human RBPs (FMR1, FXR1, FXR2, HNRNPA1, HNRNPC, IGF2BP1, IGF2BP2, IGF2BP3, KHDRBS1, LIN28B, NONO, PCBP2, PTBP1, PUM2, RBM5, QKI, RBFOX2, SF3B4, SFPQ, SRSF1, SRSF7, SRSF9, TAF15, TARDBP, TIA1 and U2AF2). The archive includes 26 files, each of which provides results for one RBP. Each file provides180-nt RNA sequences centered on the best matching sites in FASTA format, which are derived from the full-length 3’UTR. In the header/description line, the UCSC kgID, the PWM length, and the positions of the matching site in the given segment and in the full-length 3’-UTR are provided. The supportive information is added (e.g., supported_by_eCLIP_DOI_10.1038/nmeth.3810) if RBP-RNA binding is also supported by eCLIP experiments.
- beRBP-G_targets_26RBPs.7z binding targets predicted by beRBP General model.
Datasets in the manuscript
- Training and testing datasets in Table S1
- Testing datasets in Table S2~S4
- Testing datasets in Table S6
- Datasets in Figure S1