Rotation Projects

DeCiPher Project: Systematic synthetic lethal mapping of human cancer

D-Cell Project: Using deep learning to model the hierarchical structure and function of cancer cells

Computing a minimal set of genes required for life

(Updated Sept 2020) 

1. DeCiPher Project: Systematic synthetic lethal mapping of human cancer

Genetic interaction screens have been used to map gene functions and cellular pathways in model organisms and more recently in human cells. These maps can also identify therapeutic opportunities such as new synthetic lethal interactions in cancer. However, due to the requirements for large numbers of cells and associated technical challenges of these maps, most large-scale genetic interaction maps have focused on non-adherent cells, limiting the ability to study the context of these interactions in solid tumors and non-transformed epithelial tissues. Here, we engineered a large pooled combinatorial CRISPR-Cas9 library covering among tumor suppressor genes and druggable targets. To determine the context of these interactions, we mapped the genetic interaction landscape of breast cancer and lung cancer cell lines. We have also mapped these interactions in a non-cancerous, epithelial cell line to study whether these interactions are specific to cancer. Our map provides a broad resource and dissects how key cancer pathways are rewired due to cancer.  

Prerequisites: Computer programming or scripting skills.
Optional: Experimental laboratory skills, which would allow student to make tests of model predictions.

2. D-Cell Project: Using deep learning to model the hierarchical structure and function of cancer cells

Although artificial neural networks are powerful classifiers, their internal structures are hard to interpret. In the life sciences, extensive knowledge of cell biology provides an opportunity to design visible neural networks (VNNs) that couple the model’s inner workings to those of real systems. Towards this goal we have recently developed The D-Cell, a VNN embedded in the hierarchical structure of approximately 2,500 subsystems comprising a eukaryotic cell. Trained on several million genotypes, DCell accurately simulates cellular growth phenotypes and also captures the molecular pathways by which alterations to the genome lead to changes in cellular growth. This rotation project will seek to apply visible neural networks like D-Cell to develop a simulation of human cancer cells. Models will simulate the impact of patient genomes on cancer cell biology and, ultimately, tumor aggression and drug response. Datasets for building the model will be provided, along with access to tumor exomes from both public and internal sources. The goal is to determine, over a 10 week rotation, the feasibility of such a cancer model. If successful, this project could be readily developed into a PhD thesis.

Prerequisites: Computer programming or scripting skills.
ptional: Experimental laboratory skills, which would allow student to make tests of model predictions.

3. Computing a minimal set of genes required for life

A long standing question in biology is how many (and which) genes are required for life. This essential core set of genes, or minimal genome, makes up the cell’s “life support system” or “chassis and power supply” on which more complex functions and processes are built. This set of genes is of keen interest in the field Synthetic Biology, which aims to synthesize the complete minimal genome of an organism and add additional functions to this genome for biotechnological, pharmaceutical and agricultural ends. This project will attempt to use our whole-cell model of the networks and pathways in a cell to predict which genes and gene combinations are essential for life and, conversely, which genes and gene combinations can be removed. If successful, this project will be able to predict minimal genomes for synthesis and testing. It will also address whether there actually is a single “minimal genome” or whether there exist many different configurations all of which are near or at the global minimum.

Prerequisites: Computer programming or scripting skills.
Optional: Experimental laboratory skills, which would allow student to make tests of model predictions.

Software Engineers

The Ideker Lab is involved in development of several bioinformatic resources for network analysis that are widely used by the biological research community. The best known is Cytoscape, a collaborative open-source software project. Cytoscape is a leading workstation-based platform for visualizing and processing complex networks. It is widely used with approximately 17,000 downloads per month. NDEx, the Network Data Exchange, is another major project, a public web resource for sharing, storing, accessing, and publishing biological knowledge as computable networks. For more information about the Cytoscape Ecosystem & NDEx Network Cloud, visit the Software & Platforms page.

PhD Students

If you are a student who wants to be admitted to a graduate program at UC San Diego, please visit the following websites: Bioinformatics and Systems Biology (BISB)Biomedical Research (BMS)Bioengineering, and Computer Science & Engineering (CSE). If you have already been admitted to a UC San Diego graduate program, please watch Dr. Ideker’s talk at AACR 2018 , read our Research pages, and read some of our papers on the projects that interest you prior to submitting an inquiry on available positions.

Postdoctoral Fellows

The Ideker Lab is recruiting bioinformatics postdocs and senior research scientists to work on The D-Cell Project, see above for detailed project description. The open postdoctoral positions will be supported by three NIH-funded cell mapping centers: the Cancer Cell Map Initiative (CCMI), the Psychiatric Cell Map Initiative (PCMI) and the Host Pathogen Map Initiative (HPMI). Our goal is to use the data generated by these centers to develop advanced computational models of cell biology for use in translating patient data to successful therapies. Currently, there is much interest in building intelligent systems for this purpose, under the banner of precision medicine. However, the current machine learning models face particular challenges when applied to biology, including handling of extreme data heterogeneity and lack of mechanistic insight into predictions. The premise of the CCMI, PCMI and HPMI is to address these challenges by building machine learning systems that are trained to learn not only biological function but also biological structure. Further vision and details for this research are described in this Perspective article in Cell.

Applicants must have a PhD in bioinformatics, computer science or a related discipline, a strong background in machine learning and bioinformatics, and a history of productive publication.

Submit an Inquiry