Simple python script for scanning PE files based on a list of patterns and write to a csv file with three columns: sample_id, forward, reverse

prepare_sample_sheet.py

import os
import csv
import argparse

def scan_paired_end_files(dir_path, pattern_list):
    # Initialize a list to store the sample IDs and file paths
    samples = []

    # Loop through all files in the directory
    for file_name in os.listdir(dir_path):
        # Check if the file ends with any of the pattern options
        for pattern in pattern_list:
            if file_name.endswith(pattern):
                # Extract the sample ID from the file name
                sample_id = file_name[:-len(pattern)]
                # Check if both forward and reverse files exist for the current sample ID
                forward_path = os.path.join(dir_path, file_name)
                reverse_path = os.path.join(dir_path, sample_id + "_2" + pattern[2:])
                if os.path.exists(forward_path) and os.path.exists(reverse_path):
                    # Store the sample ID and file paths in the list
                    samples.append((sample_id, forward_path, reverse_path))
                else:
                    print(f"Error: Paired-end files not found for sample ID {sample_id}")
                    # break

    # Write the sample IDs and file paths to a csv file
    output_file = os.path.join(dir_path, "paired_end_files.csv")
    with open(output_file, "w", newline="") as csvfile:
        writer = csv.writer(csvfile)
        writer.writerow(["sample_id", "forward", "reverse"])
        for sample in samples:
            writer.writerow(sample)

if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Scan paired-end fastq files and generate a CSV file.")
    parser.add_argument("dir_path", help="the directory containing the fastq files")
    parser.add_argument("--pattern", nargs="*", default=["_1.non_host.fastq.gz", "_1_non_host.fastq.gz"], help="the file name pattern(s) to search for")
    args = parser.parse_args()

    scan_paired_end_files(args.dir_path, args.pattern)