Frederick A. Matsen IV

The paper introduces a simulator called antigen-prime to model influenza virus evolution and benchmarks methods for variant assignment and growth rate estimation.

The paper presents a deep learning approach that uses phylogenetic dynamic program outputs to predict edges in maximum parsimony trees, achieving high accuracy on simulated and empirical datasets.

Across 119 monoclonal germinal centers, B cell affinity rises reliably despite divergent phylogenies. Selection is noisy but persistent; mutability limits exploration. A DMS-based landscape shows saturating fitness and survivorship biases.

This study uses deep learning to detect entrenchment in antibody germline genes, revealing epistatic constraints from both intra-chain and genetically uncoupled partners.

The paper estimates neutral mutation rates and fitness effects of mutations in influenza proteins using phylogenetic analysis of over 100,000 sequences, revealing context-dependent patterns across viruses.

The paper uses simulation-based deep learning to infer the affinity-fitness response function in germinal centers from experimental data.

This paper introduces a model that separates mutation and selection processes in antibody language models to improve prediction of mutation effects on antibody function.

This paper investigates the challenges of Bayesian phylodynamic inference due to rugged tree landscapes and proposes diagnostics and strategies to improve analysis reliability.

This study examines how SARS-CoV-2 evolved to escape 1,603 antibodies by analyzing mutation dynamics and selective pressures over 3.5 years of the pandemic.

This study shows that donor T cell receptor repertoires are influenced by both HLA genetics and recent viral infections, with higher HLA heterozygosity and TCR-HLA coherence linked to increased GVHD risk.

This study compares nucleotide context models and protein language models for predicting antibody affinity maturation, finding nucleotide-based models perform better.

The paper introduces larch, a tool that explores the space of optimal phylogenetic trees to find more parsimonious trees and increase diversity in the set of optimal solutions.

The paper develops 'thrifty' models of somatic hypermutation using convolutions on 3-mer embeddings, showing they outperform 5-mer models with fewer parameters.

The paper estimates mutation rates across SARS-CoV-2 genome sites and identifies factors influencing these rates, finding significant variability and patterns related to gene boundaries, sequence context, and RNA structure.

The paper uses statistical analysis of TCR repertoire data to show microhomology influences V(D)J recombination outcomes in humans.

This study shows that higher valency immunogens promote the maturation of VRC01-class antibodies with broader neutralizing activity in animals.

TreeFormer improves phylogenetic tree reconstruction accuracy using a transformer-based tree rearrangement method, outperforming FastTree 2 on small alignments.

This study identifies cross-reactive antibodies against SARS-CoV-2 and related sarbecoviruses that target conserved epitopes and show broad neutralization potential.

The paper uses allelic inclusion data to train models that identify antibody sequence constraints, finding stronger correlations with antibody properties than existing methods.

The paper introduces SCOPE, an optics-free method to reconstruct 2D and 3D shapes using DNA barcode proximity graphs from sequencing data.

This paper identifies diverse functional antibodies targeting the S2 subunit of SARS-CoV-2 spike, including one with broad neutralization activity across multiple coronaviruses.

The paper maps how HA mutations affect neutralization by serum from different age groups, showing age-dependent antigenic effects and implications for viral evolution.

The paper uses deep mutational scanning to analyze CYP2C19 variants, revealing a tradeoff between substrate specificity and protein abundance.

The paper develops a method to identify mutations with shifted effects across DMS experiments on SARS-CoV-2 spike homologs, finding some mutations show >1,000-fold differences in impact across variants.

The paper presents methods to improve clonal family inference in B cell receptor sequencing by incorporating heavy/light chain pairing information and correcting pairing imperfections.

This paper introduces a statistical model and method for estimating cell lineage trees using maximum-likelihood phylogenetics, showing improved accuracy over existing methods in simulations and zebrafish development.

The paper introduces methods to predict antibody affinity using B cell receptor lineage structures and ancestral evolutionary information.

The paper introduces a method to infer individual immunoglobulin germline genes from B cell receptor sequences, improving accuracy over existing methods.

SANTA-SIM simulates viral sequence evolution incorporating selection, recombination, and mutation to test inference methods.