Michael J. Keiser

Learning precise segmentation of neurofibrillary tangles from rapid manual point annotations

This paper develops a deep-learning method to segment neurofibrillary tangles from brain tissue images using manual annotations, achieving high accuracy and scalability.

Network-aware self-supervised learning enables high-content phenotypic screening for genetic modifiers of neuronal activity dynamics

The paper introduces Plexus, a self-supervised model for capturing neuronal activity dynamics, and demonstrates its effectiveness in identifying genetic modifiers of neuronal activity in frontotemporal dementia.

Machine-learning convergent melanocytic morphology despite noisy archival slides

This study develops deep-learning models to detect melanocytic atypia from H&E slides using complementary IHC stains, achieving high AUROC and AUPRC scores despite limited paired tissue data.

Metric Ion Classification (MIC): A deep learning tool for assigning ions and waters in cryo-EM and x-ray crystallography structures

The paper introduces a deep learning tool called Metric Ion Classification (MIC) to classify ions and waters in cryo-EM and x-ray crystallography structures, identifying mismodeled ions in the PDB.

Evolutionary-Scale Enzymology Enables Biochemical Constant Prediction Across a Multi-Peaked Catalytic Landscape

The paper measures catalytic constants for diverse Adenylate Kinase sequences and mutants, revealing distinct catalytic peaks and challenging existing hypotheses about thermophilic enzyme kinetics.

An improved model for prediction of de novo designed proteins with diverse geometries

The paper presents a fine-tuned Alphafold2 model that better captures geometric diversity in de novo designed proteins compared to standard methods.

ChromaFactor: deconvolution of single-molecule chromatin organization with non-negative matrix factorization

ChromaFactor deconvolves single-molecule chromatin organization data using non-negative matrix factorization to identify key functional components and their molecular contributions.

Data-driven discovery of cell-type-directed network-correcting combination therapy for Alzheimer's disease

The paper identifies a cell-type-directed combination therapy for Alzheimer's disease using single-cell transcriptomics and drug databases, showing improved memory and reduced pathologies in a mouse model.

Deep phenotypic profiling of neuroactive drugs in larval zebrafish

The paper trains a deep metric learning model on zebrafish behavioral profiles to identify neuroactive compounds, achieving better performance than correlation distance and validating predictions with in vitro assays.

Learning chemical sensitivity reveals mechanisms of cellular response

The paper develops ChemProbe, a model that predicts cellular sensitivity to compounds using transcriptomes and chemical structures, and identifies genes driving ferroptosis.

In silico discovery of repetitive elements as key sequence determinants of 3D genome folding

The paper uses deep learning to assess how sequence changes affect chromatin organization, finding that repetitive elements and noncoding RNA genes significantly influence 3D genome structure.

Prioritizing virtual screening with interpretable interaction fingerprints

The paper introduces LUNA, a toolkit that generates interpretable interaction fingerprints for protein-ligand interactions to improve machine learning-based virtual screening.

Predicting Cellular Drug Sensitivity using Conditional Modulation of Gene Expression

This paper predicts cellular drug sensitivity by conditioning gene expression on small molecule structures, achieving an R2 of 0.7190.

Adding stochastic negative examples into machine learning improves molecular bioactivity prediction

This paper improves molecular bioactivity prediction by introducing stochastic negative examples during training, increasing drug-screening benchmark performance by 121.7%.

Simultaneous classification of neuroactive compounds in zebrafish

This paper introduces an open platform for classifying neuroactive compounds in zebrafish using phenotypic responses and provides a large dataset of drug effects.

Evolutionarily Conserved Roles For Blood-Brain Barrier Xenobiotic Transporters In Endogenous Steroid Partitioning And Behavior

The paper shows BBB efflux transporters regulate endogenous steroids and behaviors, and drug interactions with these transporters cause behavioral side effects.

A Simple Representation Of Three-Dimensional Molecular Structure

This paper introduces a 3D molecular representation (E3FP) that improves drug-to-target prediction compared to 2D fingerprints, with experimental validation of novel binding predictions.