Biology subjects

Korfmann, K.

Publications and source records attributed to Korfmann, K..

9 recordsLinked to original sources

bioRxiv2026

A reusable neural approach to recombination mapping for model and non-model species

Pedigree and crossing experiments can measure crossovers directly and provide the gold standard for recombination mapping, but their cost restricts fine-scale recombination mapping to only a few species. Patterns of linkage disequilibrium (LD) provide an alternative statistical approach for inferring variation in recombination along the genome. LD, however, is confounded by many evolutionary factors, such as demographic changes, life-history traits, and genomic structural variation. We present fastrho, a state-space neural-network estimator trained across a range of simulation-based priors. In simulated bottleneck and expansion scenarios, the fixed checkpoint recovered local map shape without target-specific retraining; comparisons with pyrho used lookup tables constructed under the simulation-generating history. We further evaluated generalizability across multiple species and, to account for additional confounders not represented in the initial training data, designed specialized models for inference in selfing plants, structured Arabis populations, and large- malaria-vector populations. A major biological application of the mosquito model was the construction of a five-arm recombination atlas spanning 13 Ag3 populations, providing a detailed view of recombination-rate variation across the dataset. Recombination maps inferred from Ag3 pedigrees provided independent, coarse-scale support for this atlas. Finally, analyses of resistance loci and redpoll bird supergenes demonstrate how selection and structural variation influence LD. Throughout our study, we use experimental maps for independent validation. Together, our results establish fastrho as a flexible framework for robust recombination mapping across diverse biological systems.

evolutionary biology↗

bioRxiv2026

GPU accelerated population genetics statistics using pg_gpu

Population genetics summary statistics--diversity, divergence, linkage disequilibrium, selection scans, and dimensionality reduction--are fundamental across human, agricultural, and ecological genomics. As whole-genome sequencing datasets have grown to hundreds of thousands of individuals, the cost of computing these statistics on conventional CPU implementations has become a major bottleneck: windowed scans of a single chromosome arm can take hours to days, and computation of pairwise linkage-disequilibrium statistics useful for demographic inference scales as O(n2) in sample size, often exceeding wall-clock budgets entirely. We present pg_gpu, a Python library implementing a comprehensive catalog of population-genetics summary statistics as fused CUDA kernels on NVIDIA GPUs. pg_gpu covers eleven categories spanning diversity and neutrality tests, divergence, admixture, the site-frequency spectrum, linkage disequilibrium, haplotype-based selection scans, dimensionality reduction (PCA, randomized PCA, local PCA / lostruct), distance distributions, relatedness, resampling, and a generalized weighted-SFS framework for custom{omega} estimators. On the full Ag1000G Phase 3 chromosome 3R arm (2,940 haplotypes, 10.9 million variants) pg_gpu agrees with scikit-allel and PLINK2 to machine precision while delivering a median 139x and maximum 1,096x speedup. For the multi-population LD statistics used by moments for demographic inference, pg_gpu is a drop-in replacement that yields a ~1,750-fold speedup over the native implementation. Whole chromosome arm scans, lostruct screens, and calculation of LD statistics complete on a single NVIDIA A100 in seconds to a few minutes.

evolutionary biology↗

bioRxiv2026

Fast pairwise coalescence enables gene-resolution scans for recent selection in diverse human populations

Identifying the genetic changes that shaped recent human adaptation depends on our ability to detect selection from genomic data. Summary statistics from haplotype scans have been widely used for that purpose, aggregating genetic signal over windows, though resolution is limited by linkage and their power may diminish as sweeps approach fixation, as in the case of the integrated haplotype score (iHS). Ancient DNA based scans recover signal by analysing time-series trajectories, but the majority of human populations fall outside the geographic range of any existing ancient DNA dataset. Pairwise coalescence times provide a way to complement statistics and can be applied to any modern cohort, yet computing them densely enough at cohort scale poses a computational challenge due to the quadratic growth in the number of haplotype pairs. We introduce gamma_smc_cu, a GPU implementation of the Gamma-SMC algorithm (Schweiger and Durbin, 2023) for pairwise time-to-the-most-recent-common-ancestor (TMRCA) inference. Applied to the 1000 Genomes Project (3,202 phased samples, corresponding to 6,404 haplotypes; 829,638 within-population pairs across 26 populations and five different continental ancestries; [~]1012 per-site posterior evaluations), it yields a gene-level TMRCA landscape of 17,823 autosomal protein-coding genes after masking for segmental duplications. The scan recovers well-known sweeps (LCT, SLC24A5, EDAR, FADS1, HERC2, ABCC11) and, combined with a depleted-to-enriched variant-class profile, resolves haplotype-block signals down to the gene level. Of seven case studies, two are developed in the main text -- GRK2 /ADRBK1 (chr11q13.2; SAS+EUR) and TREML1 /TREM2 (chr6p21.1) -- and the remaining five (IFIH1 chr2q24/IBS, CCDC92 chr12q24/CDX, SLC6A15 chr12q21/CHS, BPIFA2 chr20q11/GIH, CLEC6A chr12p13/CDX) are presented in the Supplementary Information (SI). Notably, TREML1 /TREM2 is a shared out-of-Africa signal -- ranked below the within-population 1% tail in 16 of 19 non-African 1000 Genomes panels that PopHumanScan and five landmark haplotype-based scans miss. A previous 10 kb-windowed-mean iHS scan dilutes the cluster of extreme sites packed inside the [~]5 kb gene bodies, while our own gene-level iHS independently recovers the locus in three South Asian panels (BEB, STU, ITU; top 0.4% genome-wide). We cross-validate the seven cases against the 9.7 million per-variant selection posteriors from a recent West-Eurasian ancient DNA scan. BPIFA2 is detected concordantly (s {approx} 1.8% per generation). GRK2 and CCDC92 reach detection threshold in flanking variants but not within their own gene bodies, while the TREML1 /TREM2 cluster falls below it. To calibrate novelty, we review the candidate landscape against an expanded eight-catalog set spanning curated haplotype scans, the largest current West-Eurasian ancient-DNA leads, and a recent 26-population iHS refinement; the vast majority of our loci overlap at least one prior entry, and only a handful -- including TREML1 /TREM2 -- remain unflagged. The contributions of this work are gene-level resolution, systematic ancient DNA cross-validation, and a reusable TMRCA landscape that complements aDNA panels.

evolutionary biology↗

bioRxiv2025

Contemporary hybridization among Arabis floodplain species creates opportunities for adaptation

Hybridization between closely related species is increasingly recognized as a major source of biodiversity. Yet, whether it can create advantageous trait combinations while purging harmful alleles remains unknown. We studied Arabis nemorensis and A. sagittata, two endangered species that currently hybridize in a single hotspot. We measured 22 phenotypic traits and mapped their genetic basis in an F2 population, after generating high quality genome assemblies for both species. In total, 58 QTLs were identified for 20 traits, with additive and dominance effects best fitting Gaussian and logistic distributions, respectively. Six large-effect QTLs were linked to significant hybrid fitness loss. Two genomic regions showed strong transmission bias favoring A. sagittata alleles, potentially accelerating their introgression. However, 48% of QTLs were unlinked to reduced fitness or segregation distortion and may generate genotypes exceeding parental performance. Notably, a major QTL affecting flowering time explained 23% of phenotypic variation and implicated TFL1 as a candidate gene for life history adaptation. While most QTLs lacked overlap with past selective sweeps, indicating limited recent positive selection, 5 of 7 QTLs for rosette size overlapped with sweep signatures in the parental lineages. Overall, our findings offer unique insights into incipient stages of hybridization.

evolutionary biology↗

bioRxiv2025

Coalescence and Translation: A Language Model for Population Genetics

Probabilistic models such as the sequentially Markovian coalescent (SMC) have long provided a powerful framework for population genetic inference, enabling reconstruction of demographic history and ancestral relationships from genomic data. However, these methods are inherently specialized, relying on predefined assumptions and/or limited scalability. Recent advances in simulation and deep learning provide an alternative approach: learning directly to generalize from synthetic genetic data to infer specific hidden evolutionary processes. Here we reframe the inference of coalescence times as a problem of translation between two biological languages: the sparse, observable patterns of mutation along the genome and the unobservable ancestral recombination graph (ARG) that gave rise to them. Inspired by large language models, we develop cxt, a decoder-only transformer that autoregressively predicts coalescent events conditioned on local mutational context. We show that cxt performs on par with state-of-the-art MCMC-based likelihood models across a broad range of demographic scenarios, including both in-distribution and out-of-distribution settings. Trained on simulations spanning the stdpopsim catalog, the model generalizes robustly and enables efficient inference at scale, producing over a million coalescence predictions in minutes. In addition cxt produces a well calibrated approximate posterior distribution of its predictions, enabling principled uncertainty quantification. We apply cxt to population genomic data from both humans and mosquitoes, highlighting the models ability to deal with the complexities of empirical data. Significance statementcxt is a language model for population genetics which introduces next-coalescence prediction as translation from observed mutations to coalescence times by modeling the coalescent with recombination as a conditional stochastic process. It learns implicit priors from stdpopsim and generalizes across both known and novel demographies. cxt generates millions of TMRCA estimates in minutes and samples well-calibrated posteriors for uncertainty quantification. A simple post-hoc correction aligns predicted diversity with the species mutation rate, ensuring robustness to novel evolutionary scenarios.

evolutionary biology↗

bioRxiv2025

Accessible, realistic genome simulation with selection using stdpopsim

Selection is a fundamental evolutionary force that shapes patterns of genetic variation across species. However, simulations incorporating realistic selection along heterogeneous genomes in complex demographic histories are challenging, limiting our ability to benchmark statistical methods aimed at detecting selection and to explore theoretical predictions. stdpopsim is a community-maintained simulation library that already provides an extensive catalog of species-specific population genetic models. Here we present a major extension to the stdpopsim framework that enables simulation of various modes of selection, including background selection, selective sweeps, and arbitrary distributions of fitness effects (DFE) acting on annotated subsets of the genome (for instance, exons). This extension maintains stdpopsims core principles of reproducibility and accessibility while adding support for species-specific genomic annotations and published DFE estimates. We demonstrate the utility of this framework by comparing methods for demographic inference, DFE estimation, and selective sweep detection across several species and scenarios. Our results demonstrate the robustness of demographic inference methods to selection on linked sites, reveal the sensitivity of DFE-inference methods to model assumptions, and show how genomic features, like recombination rate and functional sequence density, influence power to detect selective sweeps. This extension to stdpopsim provides a powerful new resource for the population genetics community to explore the interplay between selection and other evolutionary forces in a reproducible, user-friendly framework.

evolutionary biology↗

bioRxiv2024

Methylomes reveal recent evolutionary changes in populations of two plant species

Plant DNA methylation changes occur hundreds to thousands of times faster than DNA mutations and can be transmitted transgenerationally, making them useful for studying population-scale patterns in clonal or selfing species. However, a state-of-the-art approach to use them for inferring population genetic processes and demographic histories is lacking. To address this, we compare evolutionary signatures extracted from CG methylomes and genomes in Arabidopsis thaliana and Brachypodium distachyon. While methylation variants (SMPs) are less effective than single nucleotide polymorphisms (SNPs) for identifying population differentiation in A. thaliana, they can classify phenotypically divergent B. distachyon subgroups that are otherwise genetically indistinguishable. The site frequency spectra generated using methylation sites from varied genomic locations and evolutionary conservation exhibit an excess of rare alleles. Nucleotide diversity estimates were three orders of magnitude higher for methylation variants than for SNPs in both species, driven by the higher epimutation rate. Correlations between SNPs and SMPs in nucleotide diversity and allele frequencies at gene exons are weak or absent in A. thaliana, possibly because the two sources of variation reflect evolutionary forces acting at different timescales. Linkage disequilibrium quickly decays within 100 bp for methylation variants in both plant species. Finally, we developed a deep learning-based demographic inference approach. We identified recent population expansions in A. thaliana and B. distachyon using methylation variants that were not identified when using SNPs. Our study demonstrates the unique evolutionary insights methylomes provide that SNPs alone cannot reveal.

evolutionary biology↗

bioRxiv2022

Simultaneous Inference of Past Demography and Selection from the Ancestral Recombination Graph under the Beta Coalescent

The reproductive mechanism of a species is a key driver of genome evolution. The standard Wright-Fisher model for the reproduction of individuals in a population assumes that each individual produces a number of offspring negligible compared to the total population size. Yet many species of plants, invertebrates, prokaryotes or fish exhibit neutrally skewed offspring distribution or strong selection events yielding few individuals to produce a number of offspring of up to the same magnitude as the population size. As a result, the genealogy of a sample is characterized by multiple individuals (more than two) coalescing simultaneously to the same common ancestor. The current methods developed to detect such multiple merger events do not account for complex demographic scenarios or recombination, and require large sample sizes. We tackle these limitations by developing two novel and different approaches to infer multiple merger events from sequence data or the ancestral recombination graph (ARG): a sequentially Markovian coalescent (SM{beta}C) and a graph neural network (GNNcoal). We first give proof of the accuracy of our methods to estimate the multiple merger parameter and past demographic history using simulated data under the {beta}-coalescent model. Secondly, we show that our approaches can also recover the effect of positive selective sweeps along the genome. Finally, we are able to distinguish skewed offspring distribution from selection while simultaneously inferring the past variation of population size. Our findings stress the aptitude of neural networks to leverage information from the ARG for inference but also the urgent need for more accurate ARG inference approaches.

genetics↗

bioRxiv2022

Weak seed banks influence the signature and detectability of selective sweeps

Seed banking (or dormancy) is a widespread bet-hedging strategy, generating a form of population overlap, which decreases the magnitude of genetic drift. The methodological complexity of integrating this trait implies it is ignored when developing tools to detect selective sweeps. But, as dormancy lengthens the ancestral recombination graph (ARG), increasing times to fixation, it can change the genomic signatures of selection. To detect genes under positive selection in seed banking species it is important to 1) determine whether the efficacy of selection is affected, and 2) predict the patterns of nucleotide diversity at and around positively selected alleles. We present the first tree sequence-based simulation program integrating a weak seed bank to examine the dynamics and genomic footprints of beneficial alleles in a finite population. We find that seed banking does not affect the probability of fixation and confirm expectations of increased times to fixation. We also confirm earlier findings that, for strong selection, the times to fixation are not scaled by the inbreeding effective population size in the presence of seed banks, but are shorter than would be expected. As seed banking increases the effective recombination rate, footprints of sweeps appear narrower around the selected sites and due to the scaling of the ARG are detectable for longer periods of time. The developed simulation tool can be used to predict the footprints of selection and draw statistical inference of past evolutionary events in plants, invertebrates, or fungi with seed banks.

evolutionary biology↗