Koji Tsuda

pLM-bVAE: protein language model-enhanced binary representation improves quantum annealing-based protein sequence design

This paper introduces pLM-bVAE, a binary variational autoencoder enhanced with protein language model representations, showing improved sequence reconstruction and design performance in quantum annealing-based protein sequence design.

TP53 status aligns with morphology and metaplastic programs in colorectal cancer organoids, and MDM2 P2 promoter usage reflects p53 activity

This study links TP53 status to morphology and metaplastic programs in CRC organoids, showing P2 promoter usage reflects p53 activity beyond genotype in some cases.

Knowledge-guided Bayesian optimization using pre-trained LLMs speeds up the identification of superior genotypes from germplasm collection

This paper uses pre-trained LLMs to enhance Bayesian optimization for selecting superior genotypes from germplasm collections, improving efficiency with passport data and genomic information.

BacteReason: A Reasoning Model for Antimicrobial Resistance Prediction

BacteReason predicts bacterial antibiotic susceptibility with mechanistic rationales, improving accuracy over baselines using reasoning supervision.

Optimizing progeny allocation strategies in breeding schemes while updating genomic prediction models

The paper develops a framework that optimizes breeding strategies by updating genomic prediction models during breeding schemes, showing improved performance under moderate to low selection intensities.

Ab-VS: Evaluating Large Language Models for Virtual Antibody Screening via Antibody-Antigen Interaction Prediction

Ab-VS-Bench evaluates large language models for antibody virtual screening by predicting antibody-antigen interactions through three tasks: scoring, ranking, and screening.

DiffBreed: Automatic differentiation enables efficient gradient-based optimization of breeding strategies

DiffBreed uses automatic differentiation to optimize breeding strategies, achieving higher genetic gains than equal allocation through gradient-based optimization.

Prefusion conformation of SARS-CoV-2 receptor-binding domain favors interactions with human receptor ACE2

The paper compares the conformational changes in SARS-CoV-2 and SARS-CoV-1 spike proteins' receptor-binding domains, showing how the prefusion conformation of SARS-CoV-2 enhances interactions with hACE2.