janusx 1.0.21

A Suite of Genotyping Tools for Genome-Wide Association Study and Genomic Selection

JanusX

CLI Guide | Core API Guide | Zea Eureka

Overview

JanusX (Joint Association and Novel Utility for Selection) is a GWAS and genomic selection toolkit that combines:

• Rust-accelerated kernels (PyO3 extension)
• Python analysis modules
• A Rust launcher (jx) for runtime/toolchain management and pipeline orchestration

       _                      __   __
      | |                     \ \ / /
      | | __ _ _ __  _   _ ___ \ V /
  _   | |/ _` | '_ \| | | / __| > <
 | |__| | (_| | | | | |_| \__ \/ . \
  \____/ \__,_|_| |_|\__,_|___/_/ \_\ Tools for GWAS and GS
  ---------------------------------------------------------

Main capabilities:

• GWAS: LM, LMM, FastLMM, LRLMM, FarmCPU
• Genomic selection: GBLUP, adBLUP, rrBLUP, BayesA/B/Cpi, and ML models (RF/ET/GBDT/XGB/SVM/ENET)
• Streaming genotype IO for VCF/HMP/PLINK/TXT/NPY
• Post-analysis workflows: postgwas, postgarfield, postbsa
• Utility workflows: grm, pca, gformat, gmerge, hybrid, adamixture, webui, sim, simulation
• Launcher pipelines: fastq2vcf, fastq2count

---

jx & jxpy

JanusX provides two command entry styles:

• jx (Rust launcher): manages runtime/update/toolchain and supports all launcher modules
• jxpy (Python package entry): runs Python-side modules directly

Practical difference:

• fastq2vcf and fastq2count are launcher-only (jx)
• launcher-only flags (-update/-upgrade/-list/-clean/-uninstall) are not available in jxpy

---

Installation

Option A: launcher install (recommended for end users)

Download installer assets from Releases:

Then run installer:

# Linux
./JanusX-vX.Y.Z-linux-x86_64.run

# macOS
./JanusX-vX.Y.Z-darwin-universal.command

On Windows, double click or run the .exe installer.

After install:

jx -v
jx -list module

Option B: Python package install (API-first / development)

pip install janusx
# command entry: jxpy (version>=1.0.14) or jx (version<1.0.14)

• If you only need a turnkey CLI environment with pipeline tooling, prefer launcher install.

---

BLAS backend and threads

• OpenBLAS is available on macOS/Linux/Windows.
• JanusX now applies -t/--thread to BLAS thread env vars (OPENBLAS_NUM_THREADS, OMP_NUM_THREADS, etc.) in major analysis CLIs.
• For strongest cross-platform numerical/performance consistency, prefer a conda-forge OpenBLAS environment.

Example (OpenBLAS environment):

conda create -n jx_openblas -c conda-forge python=3.14 numpy scipy "libblas=*=*openblas"
conda activate jx_openblas

Default behavior is: prefer OpenBLAS, warn when unavailable, and automatically fallback to platform-priority backends.

---

Quick start

1) GWAS

jx gwas -vcf example/mouse_hs1940.vcf.gz -p example/mouse_hs1940.pheno -lmm -o test

2) Post-GWAS

jx postgwas -gwasfile test/mouse_hs1940.test0.add.lmm.tsv -manh -qq -thr 1e-6 -o testpost

3) Genomic selection

jx gs -vcf example/mouse_hs1940.vcf.gz -p example/mouse_hs1940.pheno -GBLUP -cv 5 -o testgs

* Genomic Selection for trait: test0
Train size: 1410, Test size: 530, EffSNPs: 8960
✔︎ GBLUP ...Finished [3.5s]
✔︎ adBLUP ...Finished [10.1s]
✔︎ BayesA ...Finished [32.6s]
✔︎ BayesB ...Finished [34.2s]
✔︎ BayesCpi ...Finished [32.0s]
✔︎ RF ...Finished [1m46s]
✔︎ XGB ...Finished [9m27s]
✔︎ SVM ...Finished [2m05s]
✔︎ ENET ...Finished [9.7s]
Fold Method     Pearsonr Spearmanr     R² h²/PVE time(secs)
   1 GBLUP         0.704     0.671  0.493  0.610      0.531
   1 adBLUP        0.717     0.679  0.512  0.694      2.341
   1 BayesA        0.721     0.695  0.514  0.714      5.307
   1 BayesB        0.722     0.693  0.517  0.667      5.522
   1 BayesCpi      0.699     0.671  0.476  0.672      4.971
   1 RF            0.709     0.702  0.471  0.468      3.055
   1 XGB           0.754     0.733  0.564  0.563      5.231
   1 SVM           0.703     0.664  0.490  0.485      9.027
   1 ENET          0.720     0.704  0.514  0.517      0.274

See full usages in CLI Guide.

4) Get module help

jx <module> -h

---

Module map

Genome-wide Association Studies (GWAS):

• grm
• pca
• gwas
• postgwas (Visualization, manh qq ldblock)

Genomic Selection (GS):

• gs
• reml (Estimation of broaden heritability and blup values)

GARFIELD:

• garfield
• postgarfield

Bulk Segregation Analysis (BSA):

• postbsa (Visualization, after BSA pipeline)

Pipeline and utility:

• fastq2count (RNAseq pipeline, launcher-only)
• fastq2vcf (BSA/Reseq pipeline, launcher-only)
• adamixture (Based on ADAMIXTURE)
• hybrid (Generate F1 genotype base on Parents)
• gformat (Conversion between genotype data formats)
• gmerge (Merge genotype between samples)
• webui (Visualization, beta version)

Benchmark:

• sim
• simulation

---

Citation

@article {FuJanusX,
  title = {JanusX: an integrated and high-performance platform for scalable genome-wide association studies and genomic selection},
  author = {Fu, Jingxian and Jia, Anqiang and Wang, Haiyang and Liu, Hai-Jun},
  year = {2026},
  doi = {10.64898/2026.01.20.700366},
  publisher = {Cold Spring Harbor Laboratory},
  URL = {https://www.biorxiv.org/content/early/2026/01/23/2026.01.20.700366},
  journal = {bioRxiv}
}

---

License

This project is licensed under GNU Affero General Public License v3.0 (AGPL-3.0-or-later). See LICENSE.