Hannah K. Carter

This paper introduces NEMo, a machine learning model that predicts neoantigen immunogenicity using in vivo immune editing data, showing improved stratification of progression-free survival across multiple clinical settings.

The paper evaluates how different model architectures, LD filtering thresholds, and variant prioritization methods affect gene-level eQTL score prediction quality using over 3 million models.

GenVarLoader accelerates deep learning applications in personalized genomics by improving data loading speed and compression.

NeoPrecis improves immunotherapy response prediction by integrating immunogenicity and clonality-aware neoantigen landscapes, showing better performance than tumor mutation burden in melanoma and NSCLC.

The paper identifies epistatic interactions between cancer driver mutations and allele-specific copy number variations in tumors, showing their impact on patient survival.

This paper shows the HOXB13 p.G84E variant increases prostate cancer risk by threefold in men, but does not lead to more aggressive cancer.

This paper introduces MutationProjector, an AI model that converts tumor mutation profiles into cancer subtype representations, achieving strong performance in predicting treatment responses and classifying metastasis.

GAME provides a standardized framework for benchmarking genomics models across tasks, enabling consistent evaluation and community contributions.

VADEr, a Vision Transformer-inspired framework, improves polygenic risk prediction for prostate cancer by capturing genetic interactions and revealing heterogeneity through DARTH scores.

This paper uses machine learning to predict ICB response by integrating germline and somatic features, revealing differences in immune pathways and clinical outcomes between MHC-I and MHC-II neoantigen reliant tumors.

This study identifies 165 T1D risk signals and develops a machine learning model, T1GRS, that improves risk prediction for T1D in Europeans and African Americans.

The paper uses in vitro evolution in haploid human cells to identify drug resistance mechanisms and validates them through gene expression experiments.

Bento enables subcellular analysis of spatial transcriptomics data by defining domains, annotating localization patterns, and quantifying gene colocalization, demonstrating its use on multiple datasets.

This study shows that epigenetic variants affecting DNA methylation are distributed in chromatin domains and are linked to cancer risk and prognosis.

The paper uses SEUSS to analyze the effects of 115 RUNX1 mutations on RNA profiles in myelogenous leukemia cells, categorizing them into three functional groups.

EUGENe is a FAIR Python toolkit for deep learning analysis of regulatory sequences, offering modular tools for data handling, model training, and evaluation.

The paper shows that neopeptide immunogenicity is influenced by the subcellular location of their parent proteins, improving predictions of vaccine and immunotherapy responses.

The paper develops network-based annotations for missense mutations to improve variant effect prediction, showing greater gains for somatic mutations than germline variants.

The paper examines how the N501Y mutation in SARS-CoV-2 affects MHC-II presentation, suggesting it may reduce immune response and contribute to immune escape.

The paper uses deep learning for Named Entity Recognition to predict missing metadata in SRA BioSample entries, achieving high accuracy for species, condition, and strain categories.

This study examines GPCR expression, mutations, and copy number variation in solid tumors, finding widespread differential expression and frequent mutations, suggesting potential roles as biomarkers and targets.

The paper presents a large-scale cancer pharmacogenomic screen with 11,576 experiments on 910 drug combinations across 85 cancer cell lines, enabling crowd-sourced prediction of synergistic drug pairs and biomarkers.

The paper reprocesses 250,000 human sequencing runs to extract allelic read counts for 300,000 variants, enabling variant detection across different sequencing strategies.

AmpliconArchitect reconstructs focal amplifications in cancer genomes, revealing novel insights into extrachromosomal mechanisms and oncoviral pathogenesis.

This study identifies technical artifacts in TCGA germline variant calls, showing how batch effects from DNA amplification confound germline-tumor type associations and recommends removing amplified DNA samples.