OSX与Linux下Grep差异：VCF位点提取脚本跨平台失效排查

Here are actionable improvements to make your script robust, efficient, and cross-platform compatible:

1. Use Fixed-String Matching (Eliminate Regex Ambiguity) #

Avoid relying on inconsistent regex behavior by forcing grep to treat your site strings as fixed literals with the -F flag (equivalent to fgrep). This works the same way on both BSD and GNU grep:

grep -F "$line" "$vcf"

This prevents accidental parsing of special characters (like \, ., or *) in your site strings as regex metacharacters.

2. Batch Matching (Massive Performance Boost) #

Looping through each line and running grep repeatedly is inefficient, especially for large VCF or sites files. Use grep’s -f flag to load all match patterns from your sites file in one pass—this reads the VCF only once instead of once per line:

# Keep VCF header lines (optional but recommended)
grep "^#" "$2" > "$3"
# Match all sites in a single pass and append to output
grep -F -f "$1" "$2" >> "$3"

This is orders of magnitude faster for large datasets.

3. Enforce Proper Tab Characters in Input #

Ensure your sites file uses actual tabs instead of literal \t strings. If you need to convert an existing file with \t to real tabs, use this cross-platform sed command:

sed 's/\\t/\t/g' old_sites.txt > new_sites.txt

4. Add Input Validation #

Make your script user-friendly by checking for missing arguments or non-existent files:

#!/bin/bash

if [ $# -ne 3 ]; then
    echo "Usage: $0 <sites_file> <input_vcf> <output_vcf>" >&2
    exit 1
fi

sites_file="$1"
input_vcf="$2"
output_vcf="$3"

# Verify input files exist
for file in "$sites_file" "$input_vcf"; do
    if [ ! -f "$file" ]; then
        echo "Error: File '$file' does not exist." >&2
        exit 1
    fi
done

# Run extraction
grep "^#" "$input_vcf" > "$output_vcf"
grep -F -f "$sites_file" "$input_vcf" >> "$output_vcf"

echo "Extraction complete. Output saved to $output_vcf"

5. Optional: Handle Chromosome Naming Variants #

Some VCFs use chr1 instead of Chr01 or 1—if this is a risk, normalize chromosome names in both your sites file and VCF first. For example, convert all to lowercase chr prefixes:

# Normalize sites file
sed 's/^Chr/chr/' "$sites_file" > normalized_sites.txt
# Normalize VCF and extract matches
grep "^#" "$input_vcf" > "$output_vcf"
grep -F -f normalized_sites.txt <(sed 's/^Chr/chr/' "$input_vcf") >> "$output_vcf"

内容的提问来源于stack exchange，提问作者C. John