Zhong Wang

FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean

This study evaluates FP8 quantization in GenomeOcean, finding it preserves biological fidelity while improving efficiency, but throughput gains fall short of theoretical limits.

Uncovering the Genomic Manifold via Scalable Learning from the Global Microbiome

The paper introduces GenomeOcean, a large-scale generative model trained on metagenomic data, to explore the genomic manifold hypothesis and demonstrates its alignment with another model.

Adaptive Container Service: a New Paradigm for Robust and Optimized Bioinformatics Workflow Deployment in the Cloud.

The paper introduces Adaptive Container Service (ACS), a cloud deployment framework for bioinformatics workflows that reduces failures and resource inefficiencies.

GenomeFace: a deep learning-based metagenome binner trained on 43,000 microbial genomes

GenomeFace improves metagenomic binning by combining composition and environmental abundance networks, producing more near-complete genomes faster than existing methods.

Persistent Memory as an Effective Alternative to Random Access Memory in Metagenome Assembly

This study evaluates using persistent memory to reduce out-of-memory errors in metagenome assembly, showing it can handle terabyte-sized datasets with increased scalability.

Critical Assessment of Metagenome Interpretation - the second round of challenges

The paper evaluates metagenomic software tools using realistic datasets, highlighting improvements in assembly and taxon profiling while identifying remaining challenges in virus/archaea detection and reproducibility.

Hybrid Clustering of Long and Short-read for Improved Metagenome Assembly

This paper introduces a hybrid clustering algorithm for metagenome assembly that improves accuracy and scalability by combining long and short-read clustering.

Interdependence between histone marks and steps in Pol II transcription

The paper uses machine learning to show that transcription predicts histone marks and chromatin states, and that transcription is required for maintaining active and repressive histone modifications.

Deconvolute individual genomes from metagenome sequences through short read clustering

The paper improves short read clustering for metagenome assembly by reducing under-clustering through sample-based statistics, showing effectiveness on mouse gut microbiome data.

A new method for rapid genome classification, clustering, visualization, and novel taxa discovery from metagenome

Genome Constellation improves taxa assignment and discovers novel taxa in metagenomic datasets by clustering genomes based on k-mer similarities.

SpaRC: Scalable Sequence Clustering using Apache Spark

SpaRC clusters large-scale sequence data using Apache Spark, achieving scalability and performance across transcriptomics and metagenomics datasets.

Critical Assessment of Metagenome Interpretation - a benchmark of computational metagenomics software

The paper benchmarks computational metagenomics software using complex datasets, showing strengths and limitations in assembly, binning, and taxonomic profiling.

A robust statistical framework for reconstructing genomes from metagenomic data

The paper introduces MetaBAT, a statistical framework for binning metagenomic data, identifying 173 specific genomes from a human gut microbiome dataset.

Widespread polycistronic transcripts in mushroom-forming fungi revealed by single-molecule long-read mRNA sequencing

The paper identifies widespread polycistronic transcripts in mushroom-forming fungi using long-read sequencing and shows short-read assembly is inadequate for isoform discovery.