RPFtools is a collection of scripts for the analysis of ribosome profiling data.
These tools run on Python 2.7.11 with numpy, scipy, pandas, twobitreader, and pysam.
python gtf_to_bed.py -i annotation.gtf(.gz) -a annotation.tsv > annotation.bed
produces a 12-column bed file from the gtf (e.g., from Gencode or Ensembl) with all transcripts and their ORFs (if present), as well as an annotation table containing geneIDs and gene/transcript biotypes
if only one ORF (the longest) per gene is required, this tool creates a reduced bed file with the geneID in the 3rd column
python get_longest_ORF_per_gene.py -b annotation.bed -a annotation.tsv -o annotation.collapsed.bed
python get_all_ORFs.py -b annotation.bed -G genome.2bit -s orf_stats.csv -o annotation.all_ORFs.bed
given a bed file with transcript definitions and a 2bit file with the genome, finds all ORFs of minimum length --minlength [12] and writes a (large) bed file with their coordinates. the 3rd column now contains an ID combined from transcript ID, coordinates, and a hash value computed from the ORF sequence.
orf_stats.csv is a table with statistics (ORF and UTR lengths for each ORF)
python get_RPF_phasing.py -b annotation.collapsed.bed -B RPF.bam -o RPF_phasing.pdf
given a bam file with mapped reads, collects read counts in all ORFs of the given bed file and checks for frame bias, assuming P-sites 12nt downstream of 5'end (can be changed using --offset)
python get_RPF_profiles.py -b annotation.collapsed.bed -B RPF.bam -L "29,30" -o "12,12" -N 150 > RPF_profiles.out
takes all reads of length 29 and 30nt with 12nt offset over the ORFs of the bed file and outputs for each gene the P-site density (relative to all reads in that gene) in 2x150nt windows around start and stop codon. These can then be averaged over genes for meta-gene plots.
python get_ORFscores.py -b annotation.all_ORFs.bed -B RPF.bam -L "29,30" -o "12,12" > ORFscores.out
calculates ORFscores (Bazzini et al.) for each ORF in the input bed file, using reads of length 29 and 30nt with 12nt offset. Output contains reads per million mapped over transcript and ORF, ORFscore, % of p0 positions covered and fraction of multimappers contributing to these scores. If multiple bam files are given using -B RPF_1.bam,RPF_2.bam, scores are also calculated using pooled reads (use -L "29,30|29,30" and -o "12,12|12,12" to specify lengths and offsets).
python get_RPF_counts_per_codon.py -b annotation.collapsed.bed -B RPF.bam -L "29,30" -o "12,12" -G genome.2bit > RPF_codon_counts.out
aggregates for each ORF in the input bed file all P-site counts (for A-site counts, use -o 15) over all codons. First and last codons can be excluded using -e (default: 25,1).