We are building the next generation of data science methods for curing pediatric diseases.
We are specifically focused on the following research:
- Reproducible software for processing high-dimensional microscopy readouts. We are building infrastructure to support reproducible, large-scale microscopy data processing, and developing specific tools like pycytominer, CytoTable, and CytoSnake to process large-scale microscopy images. Our aim is to improve data processing pipelines, reproducibility, data provenance, and dataset interoperability for this emerging data type.
- Microscopy representations of cell state. We derive and benchmark different approaches to extracting biologically-meaningful and reproducible representations from microscopy images. We train artificial intelligence and machine learning (AI/ML) algorithms to predict cell phenotypes from these representations. These phenotypes include cell states including various cell health and death mechanisms. Our aim is to use these representations to annotate drug screening data with phenotype and mechanism.
- Drug screening for pediatric diseases. We perform microscopy-based, in vitro drug screens to identify promising drug candidates. Our goal is to identify new therapeutic options for children with diseases like Neurofibromatosis Type 1 (NF1), neuroblastoma, and pediatric high grade glioma.
- New models of pediatric disease to aid drug screening. We develop new assays and computational methods to support finding better drugs for pediatric diseases. This includes modeling NF1 and other pediatric diseases using patient-derived organoids, developing gene-network-based targets that take advantage of polypharmacology, developing CRISPRi approaches to simulate specific high-dimensional phenotypes, and more.
How we do science
How science is performed is as important as the research topic. Science hinges on reproducibility, and our procedure maximizes both biological and computational reproducibility.
Biology is messy, but computational biology need not be!
- Open science and software. We perform all of our work in the open and release all of our software as open source. We aim to maximize the impact and reproducibility of our research by making everything we do immediately available for others to build upon.
- Scientific publishing. We submit preprints of our work and subsequently publish in peer-reviewed journals to disseminate knowledge more formally. We use pre-print review services (like Review Commons) whenever possible to improve the peer-review process. For each project, we also share project-specific github repositories (representing a lab notebook) to facilitate computational reproducibility.
We strive for creativity, integrity, inclusivity, and rigor in everything that we do.
See below for a selection of our recent papers.