Bonnie Berger

BatchRefiner improves batch integration of scRNA-seq and scATAC-seq embeddings by ensemble refinement, showing significant performance gains across datasets and metrics.

A four-network sodium acrylate/dimethylacrylamide hydrogel, cast ionic-within-ionic without neutral re-embedding, reaches about 1000-fold linear expansion and resolves individual peptide residues on standard confocal microscopes.

The paper develops a foundation model to analyze TADs and their impact on transcription, finding that TADs enhance co-regulation and this effect declines with aging and changes in cancer cells.

Promera is a generative model for biomolecular structure prediction, filtering, and design that improves binder filtering and controllable design compared to existing methods.

This paper introduces DiffEvol, a constrained diffusion framework to model evolution, and applies it to SARS-CoV-2 data to reconstruct fitness trends and vaccine-induced phase transitions.

This paper develops a new mathematical framework, ProP Gen, to model evolutionary dynamics under phenotypic uncertainty and shows it explains phenomena like phenotypic buoying and improves predictions of bacterial persistence.

The paper introduces MINT, a protein language model designed to model protein-protein interactions, outperforming existing models in tasks like binding affinity prediction and mutational effect estimation.

The paper introduces MDSINE2, a Bayesian method for inferring microbial dynamics from time-series data, and shows it outperforms existing methods in identifying ecosystem-scale interactions.

This paper introduces ProDiT, a multimodal diffusion model for generating functional proteins with multiple states, demonstrating its ability to design proteins with specific molecular functions and allosteric regulation.

The paper introduces GrID-Net, a graph neural network that detects causal locus-gene associations using cell-state parallax, uncovering noncoding mechanisms affecting 132 genes in schizophrenia.

ChronoStrain tracks low-abundance microbial strains over time with uncertainty quantification, outperforming existing methods on real and synthetic data.

This paper introduces SF-GWAS, a secure federated framework for GWAS that enables efficient and accurate analysis of private genomic data across institutions.

PHILHARMONIC infers protein interaction networks and functional organization in non-model organisms using deep learning and spectral clustering, validated in Drosophila and coral symbionts.

ralphi uses deep reinforcement learning to improve haplotype assembly, achieving lower error rates than existing methods on standard benchmarks.

This paper introduces AbMAP, a transfer learning framework for antibody sequence analysis, showing it can predict structure, binding, and paratope with high accuracy and enable antibody optimization.

SF-Relate identifies genetic relatives across distributed genomic datasets using a privacy-preserving federated approach with high accuracy and efficiency.

This paper applies parameter-efficient fine-tuning methods to proteomics, achieving competitive or better performance than traditional fine-tuning with significantly reduced computational resources.

Velorama infers causal gene regulatory networks using RNA velocity data, revealing TF interaction speeds and their links to disease.

The paper introduces new minimizer orders for large k-values using decycling sets, improving efficiency in DNA sequencing analysis.

This paper introduces a discriminative sequence model to assess transcriptomic re-identification risks by improving linking attack accuracy across datasets.

mapquik accelerates long-read mapping by using minimizer-space seeding and a pseudo-chaining algorithm, achieving high sensitivity and speed on human and maize genomes.

This paper introduces a new split-Gal4 system based on a self-excising split-intein that is fully repressible by Gal80, enabling temporal control of genetic labeling in Drosophila.

The paper develops a method to identify homologous membrane receptors in non-model corals by combining remote homology detection and functional analysis, mapping over 1000 human receptors to coral receptors.

ConPLex predicts drug-target interactions using a deep learning model that combines pre-trained protein language models with contrastive co-embedding, achieving high accuracy and scalability.

This paper presents a deep learning approach using pre-trained protein language models for drug-target interaction prediction, achieving higher accuracy and faster training than existing methods.

This paper introduces CON-Plex, a drug-target interaction model that improves specificity by distinguishing drugs from decoys using contrastive learning.

This study maps RNA translation dynamics in the human brain using ribosome profiling, identifying thousands of novel translation events and brain-specific microproteins linked to schizophrenia.

From a single sequence, OmegaFold reaches mean LDDT 0.82 on 146 CAMEO targets, between MSA-based RoseTTAFold (0.75) and AlphaFold2 (0.86), and runs about ten times faster.

This paper introduces Haystack, a method to prioritize causal transcription factors from single-cell data, validated in Drosophila, showing high accuracy in identifying drivers of differentiation.

This paper introduces SafeGENIE, a secure and efficient algorithm for LMM-based GWAS that enables collaborative analysis without data sharing.