Distinct Cell-Type-Specific Mechanisms Underlie Cognitive Dysfunction During Persistent Integrated Stress Response Activation

Kristof A. Torkenczy et al. (including Dongze He)

PNAS, in press

PNAS article in press from single-cell multi-omics work on persistent integrated stress response activation and cognitive dysfunction. Public bioRxiv preprint available.

Dynamic transitioning between MAPK-driven and WNT-driven cell states drives intestinal cancer and shapes therapy response

Amanda R. Moore et al. (including Dongze He)

Nature Genetics 58, 1331-1340 (2026)

Nature Genetics publication from my Genentech internship project, connecting WNT- and MAPK-driven colorectal cancer cell states with tumor growth and therapy response.

Promoted to Data Scientist II at Altos Labs

Advanced to Data Scientist II, focusing on foundation models and agentic scientific workflows for biological discovery.

• Developed a reference-free, tokenization-free, byte-level single-cell foundation model trained directly on raw read sequences.
• Used self-supervised pretraining with next-base language modeling, same-cell contrastive margin loss, and multi-GPU distributed training.
• Built a LangGraph multi-agent LLM system that reproduces mass-spec proteomics studies with schema-constrained extraction, self-correction, MLflow tracing, and LLM-as-judge evaluation.

QCatch: a framework for quality control assessment and analysis of single-cell sequencing data

Yuan Gao†, Dongze He†, and Rob Patro*

† Equal contribution; * Corresponding author

Bioinformatics, Volume 42, Issue 5, May 2026, btag184

A Python-based command-line tool that generates comprehensive, interactive quality control reports for single-cell quantification results from alevin-fry and simpleaf.

msproteomics — nf-core proteomics pipeline

Creator of the official nf-core end-to-end pipeline for quantitative mass spectrometry-based proteomics. Automated protein identification, quantification, differential abundance analysis, and pathway enrichment.

spatialaxe — nf-core spatial transcriptomics pipeline

Lead developer of the official nf-core pipeline for spatial transcriptomics analysis, supporting multiple technologies and data formats. Standardized workflows for cell segmentation, transcript assignment, and spatial feature extraction.

Forseti: A mechanistic and predictive model of the splicing status of scRNA-seq reads

Dongze He*, Yuan Gao, Spencer Skylar Chan, Natalia Quintana-Parrilla, and Rob Patro*

* Corresponding author

ISMB 2024

Designed the first statistical model to predict scRNA-seq read splicing status, achieving AUROC 0.9.

Joined Altos Labs as Computational Scientist I

Built scalable bioinformatics infrastructure and multi-omics analysis systems across spatial, proteomics, and single-cell programs.

• Created and maintained open-source nf-core pipelines with 30+ merged contributions across 4 pipelines and 7 reusable tool modules.
• Architected AWS S3-triggered Seqera Tower pipelines that processed 300+ Xenium slices, 250+ mass-spec experiments, and 100+ single-cell samples without manual intervention.
• Built multi-omics causal-network workflows using footprint-based activity inference and constraint-based optimization for aging and cellular reprogramming research.

Ph.D. in Biological Sciences — University of Maryland

Methods for efficient processing and comprehensive analysis of single-cell sequencing data. Advisor: Dr. Rob Patro. Concentration: Computational Biology and Bioinformatics.

• Led development of the Rust-based alevin-fry/simpleaf ecosystem for faster, memory-frugal single-cell preprocessing.
• Co-authored the Single-Cell Best Practices open-source textbook, a community standard for single-cell analysis and benchmarking.
• Designed Forseti, the first statistical model to predict scRNA-seq read splicing status, reaching AUROC 0.9.

DifferentialRegulation: a Bayesian hierarchical approach to identify differentially regulated genes

Simone Tiberi, Joël Meili, Peiying Cai, Charlotte Soneson, Dongze He, Hirak Sarkar, Alejandra Avalos-Pacheco, Rob Patro, and Mark D Robinson*

* Corresponding author

Biostatistics 25(4), 1079-1093

A novel statistical framework for identifying genes with differential regulation across conditions in single-cell RNA-seq data.

scrnaseq — nf-core single-cell RNA-seq pipeline

Core contributor to the official nf-core pipeline for single-cell RNA-seq preprocessing. Integrated the alevin-fry/simpleaf workflow as a primary analysis path alongside STARSolo, Kallisto/BUStools, and Cell Ranger.

Best practices for single-cell analysis across modalities

L Heumos et al. (44 authors including Dongze He)

Nature Reviews Genetics 24, 550–572

A comprehensive review establishing community standards for single-cell data analysis across RNA, ATAC, protein, and spatial modalities.

Deep Learning Motif Discovery for scATAC-seq — Genentech

Summer internship project that later connected to the Nature Genetics colorectal cancer publication.

• Improved a deep-learning Motif Discovery Analysis framework for scATAC-seq, enabling the first cell-type-specific MDA at Genentech.
• Discovered regulatory targets of a key transcription factor through integrated MDA and gene-regulatory-network analysis.
• Implemented the analysis workflow in Nextflow for reproducible execution.

simpleaf: A simple, flexible, and scalable framework for single-cell transcriptomics data processing

Dongze He*, and Rob Patro*

* Corresponding author

Bioinformatics 39

A simplified interface for the alevin-fry ecosystem that can process complex single-cell data types with one command, including CITE-seq and 10X feature barcoding.

Alevin-fry unlocks rapid, accurate and memory-frugal quantification of single-cell RNA-seq data

Dongze He, Mohsen Zakeri, Hirak Sarkar, Charlotte Soneson, Avi Srivastava, and Rob Patro*

* Corresponding author

Nature Methods 19, 316–322

A suite of tools achieving 2x faster performance and dramatically lower memory usage for single-cell RNA-seq preprocessing while maintaining state-of-the-art accuracy.

M.S. in Systems Biology and Bioinformatics — Case Western Reserve

Discovery of Causal Regulatory Network of System Level Measurements by Integrative Network Analysis.

B.S. in Biotechnology — Huaqiao University

Foundation in molecular biology, genetics, and biochemistry in Xiamen, China.