Steven H. Kleinstein

This paper identifies 23 candidate causal genes for critical COVID-19 by integrating severity-stratified, longitudinal cis-eQTL mapping with GWAS signals from 1,031 patients.

The paper benchmarks computational models predicting B. pertussis booster vaccine responses using multi-omics data from three cohorts, identifying key predictors of immunity and establishing a community resource for systems vaccinology.

The paper introduces metrics for assessing tensor decomposition quality in immune profiling data using a dictionary learning framework.

The paper uses scRNA-seq to analyze T cell responses in cutaneous Lyme disease, identifying CD8+ GZMK+ IFNG+ T cells as a key immune population with inflammatory properties.

AMULETY generates language model embeddings for adaptive immune receptor sequences using pre-trained models, enabling users to leverage different models for downstream applications.

The paper analyzes the epigenomic changes in innate immune cells after influenza virus infection and finds coordinated rewiring of the epigenetic landscape.

The paper introduces scCRAFT, an autoencoder for scRNA-seq batch integration that outperforms existing methods by preserving biological heterogeneity and handling unbalanced batches and rare cell types.

This study improves antigen specificity prediction by fine-tuning pre-trained antibody language models on SARS-CoV-2 and influenza antigens, showing better accuracy than models trained on embeddings alone.

The paper benchmarks 49 algorithms for predicting B. pertussis vaccination outcomes, finding that purpose-built models outperform literature-based ones due to better handling of nonlinearities and data preprocessing.

The paper identifies a multi-omics recovery factor predictive of long COVID using data from the IMPACC study, showing it can predict LC risk early after infection.

The paper introduces nipalsMCIA, a tool for multi-omics dimensionality reduction that improves speed and flexibility over existing methods for single-cell data.

This study identifies immune subtypes in hospitalized COVID-19 patients linked to clinical outcomes and long COVID risk through longitudinal multi-omics analysis.

This study investigates viral reactivation in COVID-19 patients, finding associations between reactivation of specific virus families and clinical outcomes, including mortality and PASC symptoms.

The paper introduces nf-core/airrflow, a Nextflow-based workflow for processing AIRR-seq data, validated on simulated and real datasets to analyze immune receptor repertoires.

This study identifies immune programs linked to COVID-19 severity and mortality through longitudinal multi-omics analysis of 1,152 hospitalized patients.

This paper introduces a supervised Bayesian factor model to improve multi-omics signature discovery by integrating dimensionality reduction and predictive modeling.

This study evaluates methods for inferring B cell phylogenies using paired heavy and light chain BCR sequences, finding that incorporating light chains improves accuracy and that maximum likelihood methods with separate partitions are most reliable.

This study evaluates B cell receptor sequence embeddings for predicting sequence properties and specificity, finding that BCR-specific embeddings and paired-chain sequences improve performance.

The paper evaluates computational models predicting immune responses to Pertussis Booster vaccinations using a new community resource, finding that existing models perform poorly and new models show variable performance based on features and strategies.

The paper curates a collection of >600 immune signatures from human vaccination studies and provides a data model and web dashboard for querying these signatures.

The paper introduces the Immune Signatures Data Resource, a compendium of standardized systems vaccinology datasets for comparative vaccine response analysis.

This study compares host module activation patterns and temporal dynamics across four influenza H1N1 viruses to identify conserved and differentially regulated immune responses.

The paper improves the TIgGER algorithm to accurately identify subject-specific immunoglobulin alleles from AIRR-seq data, even when alleles differ by more than 5 SNPs from known database alleles.