📘rsidImpu

A high-performance tool to annotate GWAS summary statistics with accurate rsIDs using dbSNP.

rsidImpu is designed for large-scale genome-wide association studies (GWAS), providing fast and accurate rsID matching via chromosome + position + allele comparison (with support for allele flipping and strand complement).

The tool supports QC filtering, COJO-formatted output, gzip input/output, and multi-thread acceleration.

• 15GB+ dbSNP reference files (tsv or gz)
• millions of GWAS variants
• parallel processing (OpenMP)
• gzip input support

---

✨ Key Features

✔ Accurate RSID Matching

• Matches variants by CHR + POS + allele information
• Supports:
  • Allele flipping (A1/A2 ↔ A2/A1)
  • Strand complement matching (A↔T, C↔G)

✔ High Performance

• OpenMP multithreading
• Efficient gzipped input/output
• Hash-based dbSNP lookup
• Capable of processing:
  • 15–30 GB dbSNP reference files
  • Millions to tens of millions of GWAS variants

✔ Safe for parallel execution

All components—logging, hashing, QC, matching—are thread-safe and deterministic under OpenMP.

✔ Clean & Flexible Output

• Matched rows → <out>.txt
• Unmatched rows → <out>.txt.unmatched
• GWAS alleles are never modified (A1/A2 remain as-is)
• Optional COJO output format: SNP A1 A2 freq b se p N

✔ Built-in QC Module

(Optional, user-controlled)

• Remove invalid rows (non-finite N/beta/se/freq/P)
• Filter by MAF (default: 0.01)
• Remove duplicated SNPs (keep the one with lowest P)

📦 Installation

Compile with Makefile

Requirements:

• C++11 or later
• zlib
• OpenMP (optional but recommended)

Compile manually:

git clone https://github.com/Crazzy-Rabbit/rsidImpu.git
cd rsidImpu/src
make clean
make

The binary rsidImpu will be generated in the src/ directory.

🚀 Usage

Basic example (default GWAS format)

rsidImpu
  --gwas-summary example/gwas_test_clean.txt \
  --dbsnp example/dbsnp_test.txt \
  --out example/gwas_rsid.txt \
  --dbchr CHR --dbpos POS --dbA1 REF --dbA2 ALT --dbrsid RSID

COJO Output Format Example

rsidImpu \
  --gwas-summary example/gwas_test_clean.txt \
  --dbsnp example/dbsnp_test.txt \
  --out example/gwas_rsid.cojo.gz \
  --dbchr CHR --dbpos POS --dbA1 REF --dbA2 ALT --dbrsid RSID \
  --format cojo \
  --freq Freq --beta Beta --se SE --n N --pval P

With QC enabled

rsidImpu \
  --gwas-summary example/gwas_test_clean.txt \
  --dbsnp example/dbsnp_test.txt \
  --out example/gwas_rsid.cojo.gz \
  --dbchr CHR --dbpos POS --dbA1 REF --dbA2 ALT --dbrsid RSID \
  --format cojo \
  --freq Freq --beta Beta --se SE --n N --pval P \
  --remove-dup-snp \
  --maf 0.01

Using multiple threads (OpenMP)

rsidImpu
  --gwas-summary example/gwas_test_clean.txt \
  --dbsnp example/dbsnp_test.txt \
  --out example/gwas_rsid.txt \
  --dbchr CHR --dbpos POS --dbA1 REF --dbA2 ALT --dbrsid RSID \
  --threads 16

Enable logging

rsidImpu
  --gwas-summary example/gwas_test_clean.txt \
  --dbsnp example/dbsnp_test.txt \
  --out example/gwas_rsid.txt \
  --dbchr CHR --dbpos POS --dbA1 REF --dbA2 ALT --dbrsid RSID \
  --log rsidImpu.log

Full-featured example (threads + QC + log + COJO)

rsidImpu \
  --gwas-summary example/gwas_test_clean.txt \
  --dbsnp example/dbsnp_test.txt \
  --out example/gwas_rsid.cojo.gz \
  --dbchr CHR --dbpos POS --dbA1 REF --dbA2 ALT --dbrsid RSID \
  --format cojo \
  --freq Freq --beta Beta --se SE --n N --pval P \
  --remove-dup-snp \
  --maf 0.01 \
  --threads 16 \
  --log rsidImpu.log

Show help

rsidImpu --help

---

📥 Input Formats

GWAS summary statistics

Required columns:

• CHR (or custom name)
• POS
• A1
• A2
• P (p-value)

Additional columns needed for COJO format:

• freq
• b
• se
• N

dbSNP reference

Two supported formats:

1️⃣ TSV/CSV with header

Required columns:

• CHR
• POS
• REF
• ALT
• RSID

2️⃣ PLINK .bim or .bim.gz

Automatically parsed as:

CHR  SNP  CM  POS  A1  A2

📤 Output

Matched variants

Written to: <out> or <out>.gz\

Format depends on --format:

• gwas → original GWAS columns + SNP
• cojo → SNP A1 A2 freq beta se P N

Unmatched variants

Written to: <out>.unmatched or <out>.unmatched.gz

🧪 Example Output (COJO Format)

SNP       A1  A2  freq   b    se      p       N
rs1000    A   G   0.37   0.145   0.035   1e-5    50000
rs2000    T   C   0.42  -0.080   0.025   2e-3    50000
...

🧹 QC Module Summary

--remove-dup-snp	Remove duplicated SNPs (keep lowest P)
--maf <val>	        MAF threshold (default 0.01)
Auto-QC         	Remove lines with non-finite freq/beta/se/N/P, or p outside [0,1]

🔧 Notes

• Allele matching allows:
  • swap/flip: A1/A2 ↔ A2/A1
  • strand complement: A↔T, C↔G
• Input alleles are never modified
• Only matched rows are included in the main output
• Gzip input/output is supported automatically based on filename suffix .gz

❤️ Acknowledgement

Special thanks to ChatGPT for code assistance and architectural optimization during tool development.

📫Contract

If you have any questions or suggestions, feel free to reach out:

📧 crazzy_rabbit@163.com