Michael DeGiorgio

SHADES, a deep learning framework, distinguishes evolutionary processes like neutrality, selection, and introgression using haplotype data from humans, outperforming existing methods.

The paper identifies a link between bacterial horizontal gene transfer and transposable element proliferation in nut weevils, contributing to genome expansion and molecular innovation.

The paper integrates positive selection signatures with genetic data to identify metabolically relevant loci in South Asians, finding the MAPT locus associated with insulin resistance.

This paper introduces EvoDA, a supervised learning method for predicting evolutionary models of traits, showing improved performance over traditional methods in fungal phylogenies and gene expression evolution.

This study analyzes genomic and metagenomic data from 17th-century colonial Brazil remains to determine ancestry and potential causes of death, finding European genetic links and evidence of infectious diseases.

This paper introduces PULSe, a semi-supervised method for detecting selective sweeps using positive-unlabeled learning, showing high performance across diverse genomic contexts.

This study introduces TrIdent, a transfer learning tool for detecting adaptive genomic regions, showing improved performance over existing methods in various genetic and demographic settings.

This paper investigates the consequences of gene tree-species tree mismatch in phylogenetic regression and finds high false positive rates when trees are mismatched.

The paper reveals a previously unknown genomic lineage in present-day Blackfoot Confederacy individuals and their historic ancestors, showing divergence during the Late Pleistocene.

This paper introduces T-REx, a method using tensor decomposition to extract features from haplotype images for detecting natural selection, showing high accuracy in simulated scenarios.

This paper applies time-frequency analysis to genomic summary statistics and uses convolutional neural networks to detect natural selection signals with high accuracy.

The paper analyzes ancient genomes from Northeast Brazil and compares them to other ancient and present-day data, revealing migration routes along South America's Atlantic coast and unexpected ancestry signals.

This study presents the first whole genome sequences of pre-European contact Indigenous people from Uruguay, revealing connections to ancient Panama and eastern Brazil but not modern Amazonians, suggesting a distinct Atlantic coastal migration route.

This paper derives unbiased estimators for F- and D-statistics to reduce bias from related or inbred individuals in population genetic analyses.

The paper introduces SURFDAWave, a method to identify regions of positive selection by analyzing functional data derived from genomic diversity measures, and applies it to human data to detect adaptive introgression and predict selection parameters.

This study presents whole-genome sequencing data from 4,810 Singaporeans across three ethnic groups, revealing genetic structure, admixture, and improved imputation for Asian and Oceanian populations.

The paper analyzes ancient genomes from the Pacific Northwest Coast to show genetic continuity over 10,000 years and population structure in the late Pleistocene.