Installation

SPA_SQR ships as the SPAsqr method of the GRAB command-line binary. GRAB is a pure C++17 application; all third-party dependencies are vendored in third_party/ (Eigen, Boost headers-only subset, zlib, zstd, libdeflate, plink2/pgenlib, bgen, htslib). The only requirements on your system are a C++17 compiler (g++ / clang++ / MinGW g++) and make. The build pulls nothing from the system (no apt-get, brew, vcpkg, Conan, …) and produces a single statically linked executable that runs unmodified on Linux, macOS, and Windows (MSYS2 / MinGW).

Quick install

git clone https://github.com/qhengncsu/GRAB-feat-cpp.git
cd GRAB-feat-cpp
make -j$(nproc)             # or sysctl -n hw.ncpu on macOS
./build/grab --help SPAsqr  # smoke-test

The build produces build/grab (or build/grab.exe on Windows). Copy it to anywhere on your $PATH and you’re done — there is no install step, no shared library, and no headers exposed to users.

Build matrix

Platform	x86_64	arm64 (Apple Silicon, ARM Linux)
Linux	AVX-512 / AVX2 / scalar	NEON via Eigen + scalar
macOS	AVX-512 / AVX2 / scalar	NEON via Eigen + scalar
Windows	AVX-512 / AVX2 / scalar	(untested)

The Makefile auto-detects platform (uname -s), architecture (uname -m), and AVX2 availability. GRAB_MARCH=-march=native is the default for best SIMD; override with GRAB_MARCH=-march=x86-64-v2 for a portable distribution binary.

Verifying the build

./build/grab --help
./build/grab --help SPAsqr
./build/grab --version   # build banner + linked-library versions

If --help SPAsqr prints the SPA_SQR help screen, you’re ready to move on to Workflow 1: LOCO PGS + SPA_SQR.

External tools used by the pipeline

GRAB itself has no runtime dependencies, but the surrounding pipeline calls out to two external programs that you install separately:

LDAK-KVIK (or REGENIE) — for LOCO polygenic scores in Workflow 1.
PLINK 2 — for sparse GRM construction (used in Workflow 2). Required only if you want a GRM-aware variance correction.

Both are downloaded as single binaries; no source build needed.

Note If you already have a working SPA_SQR binary (e.g., installed by a collaborator), check the version with grab --version to confirm it includes the SPAsqr method and --spasqr-mode wald support — both are required for the full workflow documented on this site.