Daniel Mas Montserrat

Multi-ancestry admixture mapping reveals ancestry-associated disease loci in the UK Biobank

This study uses multi-ancestry admixture mapping in the UK Biobank to identify ancestry-associated disease loci, finding 13 genome-wide significant associations and four novel risk loci.

High Resolution Ancestry Deconvolution for Next Generation Genomic Data

This paper introduces Gnomix, a high-resolution ancestry deconvolution method for next-generation genomic data, achieving higher accuracy and faster performance than existing methods.

ADAMIXTURE: Adaptive First-Order Optimization for Biobank-Scale Genetic Clustering

ADAMIXTURE improves genetic clustering for large datasets by combining EM with Adam optimization, achieving faster runtime and scalability without sacrificing accuracy.

Non-Parametric Ancestry Adjustment for Polygenic Scores

This paper introduces non-parametric methods to adjust polygenic scores for ancestry bias, improving accuracy for non-European populations using the UK Biobank.

Point cloud local ancestry inference (PCLAI): continuous coordinate-based ancestry along the genome

PCLAI infers continuous genetic ancestry across the genome using a point cloud representation, enabling time-stratified geographic insights from ancient samples.

Sex Checking by Zygosity Distributions

Zigo is a reference-free, ML-based method for verifying sex concordance using genotype data, achieving high accuracy across different data modalities without manual tuning.

snputils: A High-Performance Python Library for Genetic Variation and Population Structure

snputils provides a high-performance Python library for handling genetic variation and population structure data, improving efficiency and integration in genomic research.

Deep Variational Autoencoders for Population Genetics

This paper applies variational autoencoders to genomic data to capture population structure, enable dimensionality reduction, and improve data compression across diverse populations.

Analysis of a deeply-phenotyped familial hypercholesterolemia cohort from Mexico shows a role for both rare and common alleles across known dyslipidemia genes and reveals structural variation in a novel locus.

The paper analyzes a Mexican FH cohort to show rare and common alleles in FH genes and a novel locus contribute to disease risk, with replication in UK Biobank.

Polygenic risk score portability for common diseases across genetically diverse populations

This study evaluates polygenic risk score (PRS) performance across four ancestry groups for 14 diseases, finding that ancestry-specific PRS models improve accuracy and identifying cross-trait interactions.

PopGenAdapt: Semi-Supervised Domain Adaptation for Genotype-to-Phenotype Prediction in Underrepresented Populations

PopGenAdapt improves genotype-to-phenotype prediction in underrepresented populations using semi-supervised domain adaptation, outperforming supervised methods in Nigerian, Sri Lankan, and Hawaiian populations.

Machine Learning Strategies for Improved Phenotype Prediction in Underrepresented Populations

This paper introduces a machine learning toolkit to improve phenotype prediction accuracy for underrepresented populations using SNP data from diverse genomic datasets.

Neural ADMIXTURE: rapid population clustering with autoencoders

Neural ADMIXTURE uses an autoencoder to rapidly cluster large genetic datasets, reducing computation time compared to traditional methods like ADMIXTURE.

Adversarial Attacks on Genotype Sequences

The paper demonstrates how adversarial attacks can corrupt genomic analyses and explores machine learning techniques to detect such manipulations.

Deconvoluting complex correlates of COVID19 severity with local ancestry inference and viral phylodynamics: Results of a multiomic pandemic tracking strategy

The paper integrates multi-omic data to identify genetic and microbiome factors associated with severe COVID-19, independent of ancestry and social factors.

Archetypal Analysis for Population Genetics

This paper introduces Archetypal Analysis for efficient genetic clustering, showing it outperforms ADMIXTURE in speed and interpretability across human and canid datasets.

Predicting Dog Phenotypes from Genotypes

This paper predicts dog phenotypes (breed, height, weight) from genotypes using linear and non-linear methods, showing that linear methods outperform non-linear for breed classification but not for height and weight regression.

Generative Moment Matching Networks for Genotype Simulation

This paper uses Generative Moment Matching Networks to simulate SNP sequences and introduces training improvements and evaluation metrics.

XGMix: Local-Ancestry Inference With Stacked XGBoost

XGMix is a gradient boosted tree method for local ancestry inference that is accurate, simple to use, and fast to train on consumer-level hardware.