[
]()
regenie is a C++ program for whole genome regression modelling of large genome-wide association studies.
It is developed and supported by a team of scientists at the Regeneron Genetics Center.
The method has the following properties
- It works on quantitative, binary, and time-to-event traits, including binary traits with unbalanced case-control ratios and time-to-event traits with low event rates
- It can handle population structure and relatedness
- It can process multiple phenotypes at once efficiently
- It is fast and memory efficient 🔥
- For binary traits, it supports Firth logistic regression and an SPA test
- For time-to-event traits, it supports Firth cox regression
- It can perform gene/region-based tests, interaction tests and conditional analyses
- It supports the BGEN, PLINK bed/bim/fam and PLINK2 pgen/pvar/psam genetic data formats
- It is ideally suited for implementation in Apache Spark (see GLOW)
- It can be installed with Conda
Full documentation for the regenie can be found here.
Citation
Mbatchou, J., Barnard, L., Backman, J. et al. Computationally efficient whole-genome regression for quantitative and binary traits. Nat Genet 53, 1097–1103 (2021). https://doi.org/10.1038/s41588-021-00870-7
License
regenie is distributed under an MIT license.
Contact
If you have any questions about regenie please contact
If you want to submit a issue concerning the software please do so using the regenie Github repository.
Version history
Version 4.0 (New options --t2e and --eventColList for time-to-event analysis to specify time-to-event analysis and the event phenotype name, respectively; Fix algorithm used to fit logistic Firth model when using --write-null-firth to match closer to the approach used in step 2)
Version 3.6 (Bug fix for the approximate Firth test when ultra-rare variants [MAC below 50] are being tested; Address convergence failures & speed-up exact Firth by using warm starts based on null model with just covariates)
Version 3.5 (Added CHR/POS columns to snplist output file when using --write-mask-snplist; Genotype counts are now reported in the sumstats file when using --no-split; Improved efficiency of LOOCV scheme in ridge level 0; Detect carriage return in fam/psam/bim/pvar/sample files; Minor bug fixes)
Version 3.4.1 (Reduction in memory usage for LD computation when writing to text files; Fix bug rejecting valid PVAR files)
Version 3.4 (Reduction in memory usage for LD computation with dosages; Minor bug fixes for LD computation; Bug fix for when carriage returns are in optional input files)
Version 3.3 (Faster implementation of approximate Firth LRT; New strategy for approximate Firth LRT with ultra-rare variants; Relaxed convergence criterion of Firth LRT from 1E-4 to 2.5E-4)
Version 3.2.9 (Switch to robust version of ACAT to handle very small p-values; Bug fix for Step1 when sex chromosome was included in the analysis; Allow for 64 domains when using the 4-column annotation file)
Version 3.2.8 (New option --bgi to specify custom index bgi file accompagnying BGEN file; Relax matching criteria between BGEN and index bgi files to use CPRA instead of variant ID)
Version 3.2.7 (New option --force-mac-filter to apply different MAC filter to subset of SNPs; Extend maximum number of domains to 32 for 4-column anno-file; Update PGEN library)
Version 3.2.6 (Relax tolerance parameter for null unpenalized logistic regression from 1e-8 to 1e-6; Minor bug fixes)
Version 3.2.5.3 (Fix inflation issue when testing main effect of SNP in GxE model; Minor bug fixes)
Version 3.2.5 (Use pseudo-data representation algorithm as default in step 2 single variant tests; Use ACAT to get SBAT p-value across POS/NEG models; Bug fix for ACATV when set has a single variant with zero weight)
Version 3.2.4 (Relaxed the requirement on the minimum number of unique values for QTs to 3; Various bug fixes)
Version 3.2.3 (Address convergence issues in Firth regression; Various bug fixes)
Version 3.2.2 (New columns in sumstats file (N_CASES/N_CONTROLS) to output the number of cases/controls when using --af-cc; Various bug fixes)
Version 3.2.1 (New option --lovo-snplist to only consider a subset of LOVO masks; Improve efficiency of LOVO for large sets to reduce memory usage; Bug fix for SPA with numerical overflow; For SKAT/ACAT tests with Firth correction, don't include SKAT weights when running Firth on single variants)
Version 3.2 (Bug fix for SKAT/SKATO when testing on binary traits using Firth/SPA; Switched name of NNLS joint test to SBAT test altering name of corresponding options and applied Bonferroni correction before reporting its p-value [correcting for minP of 2 tests])
Version 3.1.4 (New option --par-region to specify build to determine bounds for chrX PAR regions; new option --force-qt to force QT runs for traits with fewer than 10 values [otherwise will throw an error]; phenotype imputation for missing values is now applied after RINTing when using --apply-rint; several bug fixes)
Version 3.1.2 (Reduction in memory usage for SKAT/SKATO tests; Bug fix for LOVO with SKAT/ACAT tests; Improvements for null Firth logistic algorithm to address reported convergence issues)
Version 3.1.1 (Reduction in memory usage for SKAT/SKATO tests; Improvements for logistic regressions algorithms to address reported convergence issues)
Version 3.1 (Fixed bug in SKAT/SKATO tests when applying Firth/SPA correction; Improved SPA implementation by computing both tail probabilities; New option --set-singletons to specify variants to consider as singletons for burden masks; New option --l1-phenoList to run level 1 models in Step 1 in parallel across phenotypes; Several bug fixes)
Version 3.0.3 (Skip BTs where null model fit failed; Bug fix for BURDEN-ACAT; Bug fix when nan/inf values are in phenotype/covariate file)
Version 3.0.1 (Improve ridge logistic regression in Step 1; Add compilation with Cmake)
Version 3.0 (New gene-based tests: SKAT, SKATO, ACATV, ACATO and NNLS [Non-Negative Least Square test]; New GxE and GxG interaction testing functionality; New conditional analysis functionality; see release page for minor additions)
For past releases, see here.