settings.conf

#!/bin/bash

## AutoCoEv CONFIGURATION FILE

## INPUT FILES
PROTEIN="proteins/"     # This is the **FOLDER** with proteins list(s)
SPECIES="placental.tsv"   # FILE list of species
EXTTREE="placental.nwk"   # External species tree file (e.g. from TimeTree). Specify if PhyML/PRANK need a guide.
PAIRLST=""		# A list of defined protein pairs (e.g. "negatome_mouse_pairs.tsv", needed only if: PAIRINGMANNER="defined")

## REFERENCE ORGANISM AND ORTHOLOGUES
ORGANISM="10090"	# Taxid of the reference organism (e.g. "10090" for M. musculus; "9606" for H. sapiens)
LEVEL="40674"		# Level at which to search for orthologues; 2759 (Eukaryota); 33208 (Metazoa); 7742 (Vertebrata); 32523 (Tetrapoda); 40674 (Mammalia)

## WORKING AND DATABASE DIRS 
TMP="/var/tmp/CORUM/3047.mmus"	# Working folder
DTB="/var/tmp/DB10v1"	# Folder where databases are unpacked

## THREADS UTILIZATION
THREADS="$(nproc)"	# Number of (logical) cores to use (automatically detected, but you can change it)

## BLAST OPTIONS
DETBLAST="yes"		# Detailed BLAST results ("yes", "no")
PIDENT="35.000"		# Minimum allowed idenity (%) to the reference organism
PGAPS="25"		# Maximum allowed gaps (%) to the reference organism
BLASTCORES="4"		# Number of threads to use for BLAST runs. This may or may not work. Leave blank.

# ORTHOLOGUES ASSESSMENT BY GUIDANCE
GUIDANCEMSA="muscle"	# MSA method to use (see below for options).
GUIDANCECUT="0.95"	# Sequence cutoff value, to exclude too divergent sequences (0: no cut, 0.95, 0.98)
GUIDANCEKEEP="keep"	# Shall we clean the additional results ("keep", "clean")? Not implemented yet.

## MSA OPTIONS
MSAMETHOD="prank"	# MSA method to use ("mafft", "muscle", "prank"; supported MAFFT aliaces: "mafft-linsi", "mafft-ginsi", "mafft-einsi", "mafft-fftns", "mafft-fftnsi").
MUSCLEOPTIONS=""	# Any additional options to pass to MUSCLE
MAFFTOPTIONS=""		# Any additional options to pass to MAFFT
PRANKOPTIONS=""		# Any additional options to pass to PRANK
PRANKGUIDE="noguide"	# Use external guide tree for PRANK ("exguide", "noguide")?
GBLOCKSOPT="-b5=h"	# Gblocks options, e.g. allowed gaps: "-b5=h" (half), "-b5=n" (none), "-b5=a" (all)

## PhyML OPTIONS
PHYMLOPTIONS=""		# Any additional options to pass to PhyML (e.g. "-m Blosum62")
PHYMLGUIDE="noguide"	# Use external guide tree for PhyML ("exguide", "noguide")?
TREESROOT="rooted"	# Root the generated trees by TreeBeST? ("rooted" or "noroot")

## PAIRING
PAIRINGMANNER="all"	# Pairing manner ("all" or "defined")
MINCOMMONSPCS="20"	# Minimum number of common species per protein pair
TREESCAPS="auto"	# Tree to use with CAPS ("auto", "phyml", "external")
INCR="1000"		# Divide folders of protein pairs into groups of e.g. 1000

## CAPS RUN-TIME OPTIONS
ALPHA="0.01"		# Alpha value for threshold cut-off. Do NOT leave blank
BOOT="0.6"		# Bootstrap threshold. Do NOT leave blank
CAPSOPTIONS="-c"	# Any additional options to pass to CAPS, such as Number of alignments to simulate ("-r 100"), Convergence ("-c"), or Gap ("-g 0.8") threshold (seems buggy).
REFER="-H ${ORGANISM}"	# Reference organism sequence for CAPS run, automatically set to ${ORGANISM}
PVALUE="$ALPHA"		# Post run P-value cutoff, by default equals to ALPHA

### DATABASES SECTION BELOW ###

## DATABASES version
ORTHODBVER="v101"	# Databases download version

## Databases. Names only, no ".tab" or ".gz" file extensions! 
GENEXREFALL="odb10v1_gene_xrefs"	# UniProt, ENSEMBL, NCBI, GO and InterPro ids associated with Ortho DB gene
OG2GENESALL="odb10v1_OG2genes"		# OGs to genes correspondence
ALLFASTA="odb10v1_all_fasta"		# AA sequence of the longest isoform for all genes, fasta formatted

## MD5SUMs of databases (gzipped). Change accordingly if version is different!
GENEXREFALLM5="3ab6d2efdc43ed051591514a3cc9044e" # odb10v1_gene_xrefs.tab.gz
OG2GENESALLM5="33e63fa97ee420707cd3cddcb5e282a6" # odb10v1_OG2genes.tab.gz
ALLFASTAM5="831ef830fff549857a4c8d1639a760cb"    # odb10v1_all_fasta.tab.gz