PhD student: deep learning for cancer genomics

Description

The Laboratory of Computational Biology (www.aertslab.org) is looking for a PhD student to decipher gene regulatory programs in cancer. You will engineer new AI models to predict gene expression across cancer cell states, and across different tumour types. As training data you will use in-house generated and publicly available single-cell data sets (e.g., scRNA-seq, scATAC-seq). Different deep learning strategies will be applied, including convolutional neural networks, transformers, variational autoencoders, and combinations thereof. You will also investigate how to make a pan-cancer repository of regulatory models. Depending on your background and interest, you can work closely together with our wet-lab to generate new data sets, and to experimentally validate your predictions. Further, you will use your models in collaboration with colleagues in the lab to gain mechanistic insight into cancer cell states, hence the explainability of these models (XAI) is crucial. Other applications of your models will be the interpretation of genomic DNA variation, the integration with spatial omics data, and the design of synthetic regulatory circuits as new therapeutic strategies.

This project is embedded in a Flanders-wide EOS consortium, providing interesting opportunities for collaboration with other machine learning groups (Yvan Saeys), and with experimental cancer groups (Cedric Blanpain and Jean-Christophe Marine).

Publications

Check out some of our recent publication related to deep learning and cancer cell states:

Minnoye & Taskiran, Genome Research 2020;

Kalender Atak & Taskiran, Genome Research 2021

Wouters, Nature Cell Biology 2020

Janssens, Aibar & Taskiran, Nature 2022

For all our publications, see www.aertslab.org/#publications.

Profile

• You obtained a Master in Artificial Intelligence, Bioinformatics, Computer Science, Physics, Engineering, Bio-engineering, or equivalent.
• Proficient in Python programming
• Experience with machine learning is a plus (e.g., Tensorflow/Keras/PyTorch)
• Experience with explainable AI (e.g., SHAP) is a plus
• Experience with high performance computing, software containers
• Experience with cancer genomics is a plus, but not absolutely essential
• Ability to work independently and in a team.

We offer:

• Access to state-of-the-art compute & GPU infrastructure
• A stimulating international research environment
• You can be a driving force in a new EOS consortium across three Belgian Universities
• Competitive salary and benefits
• Fully funded PhD scholarship, but encouraged to apply for a national PhD fellowships (e.g., FWO).
• Starting date: as soon as possible