Principal Scientist, Bioinformatics Human Genetics

Roche•South San Francisco, CA

About The Position

A healthier future. It’s what drives us to innovate. To continuously advance science and ensure everyone has access to the healthcare they need today and for generations to come. Creating a world where we all have more time with the people we love. That’s what makes us Genentech. Genentech’s Department of Human Genetics sits at the center of our precision medicine strategy. We combine large-scale human genetic evidence with rich molecular, cellular, and clinical data to uncover causal disease biology and translate it into actionable targets, biomarkers, and patient stratification strategies. THE OPPORTUNITY We are seeking a Principal Scientist (Bioinformatics track) to lead cutting-edge statistical genetics and AI/ML-enabled “sequence-to-phenotype” research that directly accelerates target discovery and translation—particularly in Neuroscience (e.g., Multiple Sclerosis, Parkinson’s disease, ALS). This is a methods-forward role for a scientist who thrives at the interface of human genetics, multimodal genomics, and machine learning, and who wants to see their work influence real therapeutic decisions. In this role, you will: Lead end-to-end human genetics analyses across array and sequencing cohorts, including rigorous QC, GWAS, fine-mapping, colocalization, gene prioritization, and rare-variant association (e.g., gene-based burden and variance-component approaches), with attention to multi-ancestry and bias/robustness. Integrate genetics with functional and multimodal genomics, including single-cell and multiome (RNA/ATAC), molecular QTLs, and perturbation datasets to identify causal genes, implicated cell types/states, and mechanistic hypotheses. Advance and operationalize AI/ML for genomics by evaluating and adapting modern sequence-to-function approaches for coding and noncoding variation—and critically, establishing reliability gating (calibration, uncertainty quantification, and robust benchmarking) so model outputs are decision-ready rather than purely exploratory. Translate model outputs into genetics-ready quantities (e.g., fine-mapping priors, rare-variant weights, mechanism-linked scores) that improve discovery power and interpretability while enabling systematic, repeatable locus-to-biology workflows. Anchor evaluation in experimental and real-world evidence, leveraging perturbation ground truth (e.g., reporter assays, CRISPR/base-editing studies, multiplex functional assays) and genetics/omics benchmarks to assess generalizability across cell states and datasets. Build scalable, reusable software and workflows that make advanced genetics and AI methods broadly accessible—prioritizing reproducibility, documentation, and production-quality standards. Provide scientific leadership and mentorship, partnering closely with cross-functional computational and experimental teams to shape strategy, communicate results clearly, and drive high-impact deliverables that inform research and development decisions. Maintain scientific excellence through publications, conference presentations, and contributions to strategic external collaborations. Example focus areas you may lead include: Building a neuroscience-focused variant interpretation capability that links coding and noncoding variants to molecular function and disease-relevant phenotypes, with confidence scoring and integration into downstream genetics workflows. Multiomic-first locus interpretation from fine-mapped signals to causal genes, cell states, and regulatory mechanisms Structure-guided rare-variant discovery and interpretation frameworks Genetics-informed patient stratification and biomarker development using deep phenotyping and clinically relevant outcomes

Requirements

PhD in Statistical Genetics, Computational Biology, Bioinformatics, Biostatistics, Machine Learning, Computer Science, or a related quantitative field, followed by postdoctoral experience
Demonstrated scientific impact through high-quality peer-reviewed publications and conference presentations, with evidence of intellectual leadership (e.g., first/corresponding author)
Deep expertise in human/statistical genetics spanning common- and rare-variant methods, study design, and rigorous inference
Strong statistical and programming skills (e.g., Python and/or R) and commitment to reproducible science (version control, testing, workflow management, documentation)
Experience working with large-scale genomics datasets and modern compute environments (cloud and/or HPC)
Proven ability to lead complex, ambiguous scientific problems end-to-end and deliver robust solutions
Excellent communication and collaboration skills with the ability to influence across disciplines

Nice To Haves

Demonstrated track record developing and/or scaling reusable analytics pipelines, platforms, or libraries used by multiple teams
Strong experience integrating genetics with single-cell/multiome data, molecular QTLs, functional genomics, and perturbation evidence
Applied experience with ML for genomics, including benchmarking, calibration/uncertainty, model reliability, and transfer/generalization assessment
Interest or prior experience in neuroscience, neuroimmunology, or neurodegeneration (helpful but not required)

Responsibilities

Lead end-to-end human genetics analyses across array and sequencing cohorts, including rigorous QC, GWAS, fine-mapping, colocalization, gene prioritization, and rare-variant association (e.g., gene-based burden and variance-component approaches), with attention to multi-ancestry and bias/robustness.
Integrate genetics with functional and multimodal genomics, including single-cell and multiome (RNA/ATAC), molecular QTLs, and perturbation datasets to identify causal genes, implicated cell types/states, and mechanistic hypotheses.
Advance and operationalize AI/ML for genomics by evaluating and adapting modern sequence-to-function approaches for coding and noncoding variation—and critically, establishing reliability gating (calibration, uncertainty quantification, and robust benchmarking) so model outputs are decision-ready rather than purely exploratory.
Translate model outputs into genetics-ready quantities (e.g., fine-mapping priors, rare-variant weights, mechanism-linked scores) that improve discovery power and interpretability while enabling systematic, repeatable locus-to-biology workflows.
Anchor evaluation in experimental and real-world evidence, leveraging perturbation ground truth (e.g., reporter assays, CRISPR/base-editing studies, multiplex functional assays) and genetics/omics benchmarks to assess generalizability across cell states and datasets.
Build scalable, reusable software and workflows that make advanced genetics and AI methods broadly accessible—prioritizing reproducibility, documentation, and production-quality standards.
Provide scientific leadership and mentorship, partnering closely with cross-functional computational and experimental teams to shape strategy, communicate results clearly, and drive high-impact deliverables that inform research and development decisions.
Maintain scientific excellence through publications, conference presentations, and contributions to strategic external collaborations.
Building a neuroscience-focused variant interpretation capability that links coding and noncoding variants to molecular function and disease-relevant phenotypes, with confidence scoring and integration into downstream genetics workflows.
Multiomic-first locus interpretation from fine-mapped signals to causal genes, cell states, and regulatory mechanisms
Structure-guided rare-variant discovery and interpretation frameworks
Genetics-informed patient stratification and biomarker development using deep phenotyping and clinically relevant outcomes