config.yaml

# reference genome
# where to download assembly
ref_source: http://www.genoscope.cns.fr/externe/plants/data/Boleracea_chromosomes.fasta
# name of reference fasta
ref: data/external/ref/Boleracea_chromosomes.fasta
# dict file for reference assembly
ref_dict: data/external/ref/Boleracea_chromosomes.dict
# annotations for reference assembly
gff_source: http://www.genoscope.cns.fr/externe/plants/data/Boleracea_annotation.gff
# gff file
gff: data/external/ref/Boleracea_annotation.gff
# interval list (if you want to exclude specific regions)
interval_list: data/raw/b_oleracea.interval_list

# FASTQ source
# options: 'SRA' or 'local'
# TODO: add option for 'both'
fastq_source: SRA

# if FASTQ source = SRA
# SRA project info file
# cat SraRunInfo.csv | grep oleracea > Sra_oleracea.csv
# sed -i "1s/^/$(head -n1 SraRunInfo.csv)\n/" Sra_oleracea.csv
sra_info: data/external/Sra_oleracea.csv

# sample and population ids (full list)
pop_ids: models/sample_ids.txt
# sample and population ids for (after silhouette pruning)
pruned_pop_ids: models/pruned_sample_ids.txt

# generation time for SMC++ (years)
gen: 1

# list of distinguished individuals
distinguished_individuals: models/smc/distinguished_individuals.txt

# per base per generation mutation rate for SMC++
mu: 7e-9

# directory to use as scratch space
scratch: /scratch/sdturner/

# add paths for locally installed software
# plink2
plink2: ~/software/plink2

snpable: ../software/seqbility-20091110

# msmc-tools (directory for helper scripts)
msmc_tools: ../software/msmc-tools

# filtering params
mind: 0.3