IMPUTE

IMPUTE
Name	IMPUTE
Developer	University of Oxford, Wellcome Trust Centre for Human Genetics
Released	2007
Latest release	2015
Programming language	C++
Operating system	Unix-like, macOS, Linux
License	GNU General Public License

IMPUTE

IMPUTE is a statistical imputation tool for inferring unobserved genetic variants in genotype datasets using reference panels of phased haplotypes. Developed for population-scale genomics studies, IMPUTE combines probabilistic models and reference data to predict missing single-nucleotide polymorphisms (SNPs) and short insertions/deletions across cohorts such as those genotyped on microarray platforms. The software became influential in large consortia and projects that required harmonization of disparate genotype arrays and augmentation of variant density for downstream association analyses.

Overview

IMPUTE operates at the intersection of population genetics, statistical inference, and computational biology to impute alleles at loci not directly genotyped in a study sample. It leverages phased reference haplotype panels drawn from initiatives like the International HapMap Project, the 1000 Genomes Project, and the Haplotype Reference Consortium to model linkage disequilibrium patterns observed across populations such as those sampled in the UK Biobank, the Wellcome Trust Case Control Consortium, and the Framingham Heart Study. The tool has been used alongside analysis pipelines in genome-wide association studies (GWAS), meta-analyses, and fine-mapping efforts conducted by groups including the Broad Institute, the Wellcome Centre for Human Genetics, and academic centers at Harvard Medical School, Stanford University, and the University of Cambridge.

Methodology

IMPUTE implements a hidden Markov model (HMM) framework to model the switching among haplotype states along chromosomes, drawing conceptual lineage from methods used in coalescent theory and ancestral recombination graphs developed in theoretical work at institutions like Princeton University and University College London. It uses phased reference panels produced by phasing algorithms such as SHAPEIT and Beagle and integrates recombination maps from projects like the HapMap Recombination Map and the deCODE recombination map. The core algorithm assigns posterior probabilities to genotype configurations at untyped loci by marginalizing over possible haplotype pairings; this probabilistic inference shares foundations with expectation-maximization schemes used in statistical genetics research at the Broad Institute and the Wellcome Trust. Parameters such as effective population size and mutation models are often tuned using population genetic summaries estimated from datasets like the 1000 Genomes Project and the Genome Aggregation Database initiatives.

Applications

IMPUTE has been applied in diverse human genetics studies including GWAS for complex traits and diseases investigated by consortia like the Psychiatric Genomics Consortium, the CARDIoGRAMplusC4D Consortium, and the DIAGRAM Consortium. It is used in fine-mapping to refine associations near loci identified by the Genetic Investigation of ANthropometric Traits (GIANT) consortium and in Mendelian randomization analyses informing studies at the National Institutes of Health and the European Research Council. Population genetics studies involving samples from the Human Genome Diversity Project, the Simons Genome Diversity Project, and the African Genome Variation Project have employed IMPUTE to harmonize variant sets for comparative analyses. Public health genomics programs such as the All of Us Research Program and the Estonian Biobank have incorporated IMPUTE into genotype processing workflows to increase imputation quality for diverse ancestries.

Performance and Accuracy

Imputation accuracy of IMPUTE depends on reference panel size and ancestry match; large panels like the Haplotype Reference Consortium and the multi-ancestry 1000 Genomes Project generally yield higher concordance and imputation info scores used by GWAS analysts at institutions such as Johns Hopkins University and the University of Michigan. Benchmarking studies conducted by groups at the Wellcome Trust, the Broad Institute, and academic centers in Europe have compared IMPUTE against contemporaries such as MaCH, minimac, and Beagle, reporting competitive accuracy especially for low-frequency variants when dense reference panels and high-quality phasing are available. Computational performance scales with sample size and marker density; high-performance computing resources at centers like the European Bioinformatics Institute, the San Diego Supercomputer Center, and national supercomputing facilities are commonly used to run chromosome-wide imputation jobs.

Software Implementations

IMPUTE has been released in multiple major versions with implementation in C++ and accompanying utilities for input/output in formats compatible with PLINK, VCF, and BGEN used by projects including the UK Biobank and gnomAD. Workflow integration typically involves pre-phasing with SHAPEIT or EAGLE, conversion via tools maintained by the Sanger Institute and the Broad Institute, and post-imputation quality control using packages such as QCTOOL and SNPTEST. Containerized and pipeline implementations have been provided by bioinformatics groups at the European Molecular Biology Laboratory, the Wellcome Sanger Institute, and cloud platforms supported by the National Center for Biotechnology Information and Amazon Web Services.

Limitations and Challenges

IMPUTE’s accuracy is limited by reference panel composition, population stratification present in cohorts like those studied by the International Cancer Genome Consortium, and phasing errors introduced by upstream tools such as SHAPEIT when applied to admixed populations sampled by projects like PAGE. Rare variant imputation remains challenging compared to sequencing approaches used by the Centers for Disease Control and Prevention and the National Human Genome Research Institute, and structural variants are not comprehensively imputed by older versions. Computational demands pose barriers for resource-limited groups, prompting shifts toward faster imputation engines like minimac4 and algorithmic improvements led by teams at universities such as Massachusetts Institute of Technology and Stanford.

History and Development

IMPUTE was developed in the mid-2000s by researchers at the Wellcome Trust Centre for Human Genetics, influenced by earlier statistical genetics work from groups at the Massachusetts Institute of Technology, University of Oxford, and University College London. Major updates coincided with releases of the HapMap Project and the 1000 Genomes Project reference panels and with methodological advances from collaborators at the Broad Institute and the Sanger Institute. Subsequent versions incorporated refinements in HMM design, support for larger reference panels, and improved compatibility with emerging genotype file formats standardized by consortia such as the Global Alliance for Genomics and Health and the International HapMap Consortium. Category:Genetics software