DBIS

Project Background

Cancer heterogeneity is a major factor driving differences in treatment response and drug resistance. The MOSAIC project aims to build the first cross-cancer malignant cell atlas using single-cell transcriptomics (scRNA-seq) and advanced deep learning foundation models.

Our goal is to develop large-scale self-supervised AI models (based on Transformers and Graph Neural Networks) to tackle key biological and clinical tasks, such as:

• Cell type annotation
• Gene regulatory network inference
• Perturbation-response prediction
• Clinical phenotype prediction

This is an interdisciplinary collaboration between experts in computer science, artificial intelligence, biology and medicine at RWTH Aachen University.

Responsibilities

• Contribute to the development and implementation of deep learning models (e.g., Transformers, GNNs) in PyTorch.
• Assist in preprocessing and standardizing single-cell transcriptomics datasets (including normalization and batch-effect correction).
• Support the design of large-scale model pretraining and fine-tuning pipelines.
• Collaborate with computer scientists and biomedical researchers to apply models in real-world biological contexts.

What We Offer

• Opportunity to work in an interdisciplinary and international research team (Computer Science + Biomedical Research).
• Hands-on experience with large-scale biological datasets and state-of-the-art AI methods.
• Flexible working hours (approx. 14h/week student assistant contract).
• Possibility to contribute to scientific publications and project outcomes.

How to Apply

Please send your CV (in English or German), the transcript of current grades, and a short motivation letter to:

• Dr. Sikander Hayat: shayat@ukaachen.de
• CC Yongli Mou, M.Sc.: mou@dbis.rwth-aachen.de