config.txt

[all]
# output directory path for all created files. Recommended to use the 'build' directory in this repository.
Outdir= /path/to/ORF-Annotation-main/build/

# Define a path to your McGillivray et al. 2018 published catalog (doi: 10.1093/nar/gky188, Git: http://github.gersteinlab.org/uORFs/). Must be placed in 'build' directory.
McGillivray_path= /path/to/ORF-Annotation-main/build/complete_uORF_predictions_hg19.tsv

# Define a path to an Ensembl .gff3 file (Ensembl release 87 or higher preferred)
# can be downloaded from http://ftp.ensembl.org/pub/grch37/current/gff3/homo_sapiens/Homo_sapiens.GRCh37.87.chr.gff3.gz. Must be placed in the 'build' directory.
gff3_path=/path/to/ORF-Annotation-main/build/Homo_sapiens.GRCh37.87.chr.gff3

# Define a path to a reference FASTA (.fa) file - this is used to check that site identities match a FASTA reference build and must be placed in the 'build' directory of the repository.
# this is required for create_range_df and create_range_df scripts.
FASTA_path = /path/to/ORF-Annotation-main/build/GRCh37_latest_genomic.fna

# Subset to RefSeq definitions only (using Ensembl:RefSeq matched transcripts download from Biomart) – please set either “Yes” or “No.” If yes, results in a smaller catalogue with stricter transcript definitions.
refseq_only = Yes

# path to an Ensembl:RefSeq GRCh37 biomart file - can use the example file in the 'build' directory
# must have 'Transcript stable ID' and 'RefSeq mRNA ID' columns if you download your own biomart file
refseq_path = /path/to/ORF-Annotation-main/build/GRCh37_biomart_refseq_export.txt