01_gencall_idat_to_gtc.pbs

#!/bin/bash

#SBATCH --output=/mnt/p2000/data4/projects/catarina/illumina_gda/intermediate/01_gencall_idat_to_gtc/error/idat_to_gtc.log
#SBATCH --error=/mnt/p2000/data4/projects/catarina/illumina_gda/intermediate/01_gencall_idat_to_gtc/error/idat_to_gtc.err
#SBATCH --job-name=idat_to_gtc
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=4
#SBATCH --mem=24G
#SBATCH --partition=hc
#SBATCH --nodelist=node01

# Configura bibliotecas para uso do conda
source /opt/miniconda3/etc/profile.d/conda.sh

# Ativa ambiente conda com R 3.6.1
conda activate bio

# Configura variaveis
bpm_manifest_file="/mnt/p2000/home/catarina/illumina_gda/raw/nov_2022_LTO/diag/GDA-8v1-0_D1.bpm"
egt_cluster_file="/mnt/p2000/home/catarina/illumina_gda/raw/nov_2022_LTO/diag/GDA-8v1-0_D1_ClusterFile.egt"
path_to_output_folder="/mnt/p2000/data4/projects/catarina/illumina_gda/intermediate/01_gencall_idat_to_gtc/"
base_directory="/mnt/p2000/home/catarina/illumina_gda/raw/nov_2022_LTO/"

LANG="en_US.UTF-8" 

cd /mnt/p2000/data4/projects/catarina/illumina_gda/scripts/logs/
 
# Use find to get a list of directories (excluding "diag")
find "$base_directory" -maxdepth 1 -type d \( ! -path "$base_directory" -a ! -name "diag" \) -print0 |
while IFS= read -r -d $'\0' path_to_idat_folder; do
     # Call iaap-cli gencall for each subdirectory
     iaap-cli gencall \
        "$bpm_manifest_file" \
         "$egt_cluster_file" \
        "$path_to_output_folder" \
         --idat-folder "$path_to_idat_folder" \
         --output-gtc \
         --gender-estimate-call-rate-threshold -0.1
done