Flora Jay

DAIseg is a new HMM-based method that detects both Neanderthal and Denisovan introgression in modern human genomes without archaic references, outperforming existing methods in Papuan populations.

The paper develops a multi-task neural network to predict blood metabolite profiles by separating covariate, genetic, and joint effects, finding nonlinear covariate modeling improves prediction.

TRACTINATOR infers continental and archaic ancestry in admixed genomes using deep learning, revealing archaic introgression patterns in Latin American populations.

This paper shows that ignoring gene-gene interactions in GWAS can lead to false positive results, highlighting the need for more complex models.

This paper investigates using imputation to improve archaic ancestry inference in ancient human genomes, showing that imputed genomes can reliably detect and quantify introgression and identify introgressed segments.

Attentive-SPIDNA uses attention mechanisms to infer population size history from genomic data, outperforming existing methods and handling varying haplotype numbers.

This paper evaluates neural network-based methods for local ancestry inference in challenging admixture scenarios and finds they perform best with specialized preprocessing.

This study identifies ancient S. enterica Paratyphi C DNA in 19th-century Mexico, linking it to European strains and showing its persistence in urban settings.

This study uses 30 new palaeogenomes to show that Neolithic expansion into Western Anatolia involved both indigenous assimilation and eastern population influx, with cultural diversity decoupled from genetic differences.

This study evaluates simulation-based supervised machine learning methods for inferring demographic parameters from genomic data, showing MLP outperforms other ML and ABC methods.

The paper introduces MICFiller, a Denoising AutoEncoder to impute missing MIC values for Pseudomonas aeruginosa, achieving high accuracy compared to existing methods.

This study introduces a frugal diffusion model for generating artificial genomes and evaluates its utility for local ancestry inference using synthetic and real data.

The paper generates synthetic genomes using PCA and GANs, showing improved population structure and privacy metrics compared to existing methods.

This paper introduces convolutional and conditional neural networks for generating large-scale artificial genomes with high SNP numbers while minimizing privacy leakage.

This study benchmarks haplotype inference methods on ancient DNA by simulating aDNA data and shows that accurate phasing is possible even for very ancient individuals with sufficient contamination control and read depth.

The paper analyzes pre-Hispanic genetic data from Mexico to show population continuity despite climate change and identifies an ancient unsampled ghost population's contribution.

This paper shows that cultural transmission of reproductive success affects genomic diversity, coalescent tree topologies, and demographic inferences by altering effective population size and distorting tree structures.

This study uses generative models to create artificial human genomes that can improve genetic research without compromising privacy.