Postdoctoral Research Associate in Statistical Genetics

Project

The postdoctoral associate will develop statistical methods for mapping cell‑type‑specific expression quantitative trait loci (eQTLs) from bulk RNA‑seq cohorts, using single‑cell RNA‑seq data as a reference. The work builds on the lab’s Bayesian deconvolution framework, BayesPrism (Nature Cancer, 2022), and extends it to a joint statistical model of the single‑cell reference, bulk expression, and genotype. In parallel, the lab is developing deep generative models for statistical deconvolution, and the associate will have the opportunity to work on this direction as well. The methods will be applied to large‑scale bulk and single‑cell transcriptomic datasets with matched genotypes, and the resulting cell‑type‑specific eQTLs will be integrated with GWAS. The successful candidate will lead this project.

Responsibilities

• Develop and implement statistical models and inference procedures for cell‑type‑specific eQTL mapping.
• Validate models through simulation studies and held‑out data.
• Apply methods to study cohorts in collaboration with faculty across the Department of Genome Sciences and the University.
• Prepare manuscripts for peer‑reviewed journals and present findings at scientific meetings.
• Release documented software for public use.
• Mentorship: receive individualized mentorship focused on algorithm and model development, scientific writing, and grant preparation; encouraged to develop independent research directions.
• Opportunities to receive hands‑on training in deep generative modeling and to apply for independent fellowships.

Qualifications

• Ph.D. in Statistics, Biostatistics, or a related quantitative field, completed by the start date.
• Demonstrated experience developing statistical methodology, especially in statistical inference, evidenced by a first‑author methods paper (published or accepted) that includes the candidate’s own derivation and implementation.
• Working knowledge of hierarchical Bayesian models, mixed models, latent‑variable models, or high‑dimensional inference.
• Proficiency in R or Python.
• Experience with eQTL or GWAS analysis is desirable but not strictly required.

Schedule

• 12‑month appointment, renewable contingent on satisfactory performance and the availability of funding.