Research

My research develops statistical and machine learning methods for computational biology with a focus on mass spectrometry-based proteomics, measurement-aware statistical modeling, and causal inference for biological systems. Across these areas, I build methods that account for the realities of biological experiments: complex measurement processes, limited replication, missing data, and variable measurement quality.

I aim to develop methods that are both statistically rigorous and practically useful. In addition to methodological research, I translate these ideas into open-source software and collaborative workflows that support researchers in academia and industry.

Statistical methods and software for proteomics

A major part of my research focuses on developing statistical methods for quantitative proteomics experiments, especially mass spectrometry-based assays with complex designs and heterogeneous sources of variability. These experiments often require specialized models that account for the structure of peptide- and protein-level measurements, missingness, experimental design, and assay-specific confounding.

I have developed methods for multiple classes of proteomics experiments, including post-translational modification analysis, limited proteolysis experiments, and scalable workflows for large data-independent acquisition studies. I am also a lead developer and maintainer of the MSstats ecosystem, a collection of open-source Bioconductor tools for statistical analysis of proteomics experiments. These tools are used across academic and pharmaceutical laboratories and are designed to make rigorous statistical workflows accessible, reproducible, and extensible.

Measurement quality-aware statistical modeling

Modern large-scale proteomics experiments often contain substantial variation in measurement quality across runs, features, and conditions. Most downstream statistical methods treat all summarized measurements as equally reliable, even when some observations are clearly less trustworthy because of poor chromatographic behavior, noisy peaks, or other acquisition artifacts.

My recent work develops quality-aware statistical methods that explicitly model measurement reliability and propagate this information into downstream protein-level inference. In particular, I developed the MSstats+ framework, which integrates anomaly detection and longitudinal quality metrics into differential analysis for DIA proteomics. These methods allow high-quality observations to contribute more strongly while reducing the influence of poorly quantified measurements.

This work reflects a broader research interest in models that account for how biological data are generated, not just the final summarized measurements. I am especially interested in statistical approaches that connect experimental quality control with downstream inferential robustness.

Causal modeling for biological systems

Beyond proteomics-specific methodology, I am developing causal modeling frameworks for biological systems that integrate molecular measurements with prior biological knowledge. These approaches are aimed at estimating the effects of perturbations, predicting responses to new interventions, and improving inference in settings with sparse replication and incomplete experimental coverage.

My current work in this area combines proteomics, transcriptomics, structured pathway knowledge, and Bayesian modeling to support interventional prediction in molecular systems. This includes the software platform Causomic, which integrates biological knowledge resources with causal inference models for systems-level analysis.

Looking ahead, I am interested in building context-aware and multi-omics causal models that can represent how regulatory relationships vary across tissues, conditions, and molecular modalities. More broadly, my long-term goal is to develop machine learning and statistical frameworks that support iterative, data-driven modeling of cellular systems.

Collaborative and translational focus

My research is shaped by close collaboration with experimental scientists, method developers, and pharmaceutical researchers. I have worked with collaborators at Genentech, Pfizer, AstraZeneca, and Talus Bio, where methodological questions arise directly from real experimental workflows and translational research settings. These collaborations help ground my research in practical scientific problems while motivating new methodological directions.

Selected themes

• Statistical methods for quantitative proteomics
• Measurement quality-aware inference
• Causal modeling of biological systems
• Multi-omics integration
• Open-source scientific software