example.yml

# Where to store results
output_directory: 'out'
# Human annotation file from GENCODE (gff3)
gencode: "/path/to/gencode.v33.primary_assembly.annotation.gff3"
# CAT annotation to filter
cat_gff3: "/path/to/species_CAT_annotation.gff3"
# (bgzipped or not) FASTA file for species with CAT annotation
cat_fasta: "/path/to/species_CAT.fa.gz"
# (bgzipped or not) FASTA file for hg38
hg38_fasta: "/path/to/hg38.fa"
# Name of mitochondrial sequence in CAT genome
cat_mito: "chrM"
# Use synteny to filter genes? If false, only discards blacklisted
# gene/transcript types
synteny: true
# ===== OPTIONAL =====
# Optionally filter a human annotation (e.g. 10X) to only 
# include genes and transcripts in the filtered list, and 
# to use the HGNC-preferred names. 
# human_annotation: "genes.gtf.gz"
# ===== OPTIONAL: rescue genes from an Ensembl annotation
# Attempt to rescue removed genes from an Ensembl annotation? 
scavenge_ens: true
# ID of genome in Ensembl. Should be like this: hsapiens for Homo sapiens
ens_id: "mmulatta"
# Ensembl GTF/GFF to scavenge (can be gzipped)
ens_gtf: "/path/to/species_ensembl.gtf.gz" 
# If the Ensembl annotation corresponds to the same assembly you're using
# (in CAT), provide a FASTA file for the annotation
ens_fasta: "/path/to/species_ensembl.fasta.gz"
# If the Ensembl annotation corresponds to a different version assembly
# for the same organism you're using (in CAT), provide a UCSC chain file
# to lift from the Ensembl version to the current (CAT) version
chain: null
# If the Ensembl annotation corresponds to a different version assembly
# for the same organism and you are providing a UCSC chain file to list
# from the old/Ensembl version to the new version, you must provide a 
# FASTA for the UCSC versioned assembly here
ucsc_fasta: null