Haiyuan Yu

The paper constructs a 3D structural interactome of SARS-CoV-2-human protein interactions, analyzing evolutionary divergence, population variation, and drug impacts.

The paper evaluates the functional impact of missense variants in human reproductive genes using mouse models and finds that many predicted pathogenic variants do not cause detectable phenotypes.

The paper identifies MLH1/3 variants in mice that cause reproductive defects, including subfertility and increased pregnancy loss due to meiotic abnormalities.

This study uses FRB-FKBP dimerization to disrupt Mlh1-Pms1 IDRs, revealing their role in DNA clamping during mismatch repair.

This study links SARS-CoV-2 infection to brain microvascular injury and neuroinflammation in dementia-like cognitive impairment through network-based genomics analysis.

This study shows that PGRN deficiency affects amyloid-beta levels and microglial activation in a sex- and age-dependent manner, influencing lysosomal activity and inflammation in Alzheimer's disease.

The paper identifies two RABL2A variants causing male infertility and ciliopathies by combining in silico predictions, protein stability assays, and mouse modeling.

The paper shows that structure-based validation underestimates error rates in proteome-wide XL-MS and proposes four metrics for better quality assessment.

The paper identifies fundamental enhancer units defined by transcription start sites and shows adjacent units operate non-cooperatively.

PIVOTAL prioritizes variants of uncertain significance using 3D proteome spatial patterns, predicting pathogenicity for 140,000 VUS and providing a web tool for analysis.

MaXLinker improves cross-link identification by reducing false positives and increasing sensitivity, revealing 9,319 cross-links in K562 cells.

GRAM predicts the molecular effect of non-coding variants in a cell-type specific manner using SELEX and expression profiles, achieving high performance on MPRA and luciferase assays.

This paper predicts adverse drug-drug interactions using genetic interaction features and a gradient boosting classifier, identifying 432 novel interactions and providing mechanistic insights.

This study predicts drug targets using a class imbalance-tolerant machine learning framework, identifying new drug-target associations for potential cancer treatment.

The paper introduces Interactome INSIDER, a browser for genomic studies that provides predicted protein interaction interfaces and enables analysis of mutations in these interfaces.

GeMSTONE is a cloud-based tool for prioritizing germline mutations from whole-genome/exome sequencing data, offering customizable workflows and reproducibility through recipe files.