1 impasse des ateliers
94400 Vitry Sur Seine
The purpose of this project would to benchmark classification methods for single cell data. The recent droplet based techniques produce high throughput , but also noisier data than the first generation of methods. In this context it is important to understand how this impacts the power of cell type classification methods, and especially methods that are based on signatures. We will compare a series of publicly available methods ranging from MCP Counter, GSVA and going to more sophisticated classification methods like Garnett and ACTINN. The final objective is to produce a series of best practices that could be applied to analyzing internally generated data.
Desired competencies: programming (R prefereable), statistics, bioinformatics, machine learning, immunology