David M. McCandlish

The paper develops a framework to analyze long-term sequence dynamics from context-dependent mutation processes, applied to somatic hypermutation in antibodies, revealing mutational biases and their effects on sequence evolution.

GaugeFixer is a Python package that efficiently fixes gauge freedoms in sequence-function models, enabling faster and more scalable analysis of large parameter sets.

This paper introduces a framework to infer fitness landscapes with structured epistasis and applies it to protein and RNA mutagenesis data, revealing complex genetic interactions.

The paper establishes the relationship between Gaussian process priors and regularized regression in sequence-to-function mapping, showing how to construct regularizers that correspond to explicit Gaussian process priors and characterize implicit function space priors.

This study introduces pVEP, a framework showing how genetic background affects AI predictions of variant effects, revealing variant effects vary across haplotypes.

The paper introduces weakest-link epistasis as a framework to analyze genetic load and shows how increasing phenotypic dimensionality affects fitness, comparing it to Fisher's geometric model.

This paper introduces SEAM, a deep learning framework that identifies functional binding sites and mutation impacts in regulatory DNA, revealing two classes of regulatory signals in human and Drosophila loci.

The paper quantifies how taxon-specific mutational biases influence adaptive amino acid substitutions across 14 species.

The paper introduces interpretable Gaussian process models for sequence-function relationships, capturing epistasis and achieving superior predictive performance on large biological datasets.

This study compares deep mutational scanning and CRISPR base editing for variant function annotation, finding strong correlation between base editor data and gold standard DMS results.

The paper identifies cryptic genetic variants in tomato that drive phenotypic change through hierarchical epistasis in a gene regulatory network.

The paper finds that positive genetic interactions have greater scientific impact but are underrepresented in literature, despite equal detection capability.

This paper analyzes gauge freedoms in sequence-function models and shows their relationship to symmetry and parameter interpretability.

This paper introduces a family of gauges for sequence-function models to remove gauge freedoms and improve model interpretation.

gpmap-tools uses Gaussian processes to infer and visualize genotype-phenotype maps with higher-order epistasis; across 262,144 Shine-Dalgarno variants it reveals register-specific peaks linked by three-nucleotide-offset ridges.

The paper analyzes over 350,000 amino acid replacements to show that fitness effects follow a maximum entropy distribution, with implications for mutational prediction and protein design.

This paper introduces SQUID, a framework for interpreting genomic deep neural networks using surrogate models, showing improved motif consistency and variant-effect predictions.

MAVE-NN learns genotype-phenotype maps from MAVE data by using a neural network-based information-theoretic framework, enabling biophysically interpretable models.

The paper introduces a non-parametric Bayesian method for density estimation in sequence space and applies it to study pre-mRNA splicing and karyotypic diversity in human cancer.

The paper introduces a method to impute sequence-function relationships by minimizing epistasis, showing it provides a less epistatic fit with comparable predictive power.

This paper introduces two new Bayesian prior methods for calibrating phylogenetic trees using fossil data and shows they improve divergence time accuracy.