PhD or Postdoc: multimodal machine learning for molecular maps of human tissue

There is a position available for either a PhD candidate or a postdoctoral researcher in TU Delft's computational team of HuBMAP: the international Human BioMolecular Atlas Program funded by the US National Institutes of Health (NIH). Our team at the TU Delft is one of three participating European partners amongst all 50 funded institutions.

The HuBMAP project: Deciphering the functioning of organs and tissues is crucial to advance our understanding of human health, as well as the diagnosis and treatment of disease. Such intricate biological functions result from the specialization, interaction, and spatial organization of all cells into higher-order structures, tissues, and organs. Scientists estimate there are 37 trillion cells in an adult human body, each with its own rich molecular profile. Determining function and relations among these cells is therefore a significant undertaking requiring the development of advanced molecular biology technologies and computational methods. The goal of HuBMAP is to drive the development of this methodology to molecularly map healthy cells in human tissue.

Your role: You will do research on multimodal machine learning methodology to discover and leverage patterns across heterogeneous high-throughput molecular profiles. Some topics of interest:

• Integration of measurements for different types of molecular species (using techniques such as transcriptomics, proteomics, and lipidomics).
• Multiscale analysis along various dimensions at different resolutions, spatial and temporal.
• recognition of coherent and contrasting/differential patterns.
• Supervised learning of cell types and tissue structures based on molecular content.
• Interpretability of multimodal pattern recognition and prediction models.

Your embedding and collaborations: This position is in the Gonçalves lab at the TU Delft, whose research focuses on algorithmics and machine learning for precision biomedicine. We are part of the Pattern Recognition and Bioinformatics section of the Intelligent Systems department, providing direct and abundant access to experts in the field of machine learning. We actively collaborate with biomedical research teams on various topics, and are also part of initiatives like Convergence for Health and Technology. Within HuBMAP, our team works in close collaboration with the Spraggins lab at Vanderbilt University (U.S.), developing experimental protocols and generating the molecular profiles of human tissue. This will enable you to gain hands-on experience with unique and exciting data from the latest technologies, including sequencing-based spatial transcriptomics and imaging mass spectrometry. In this project we also collaborate with the Van de Plas lab at the TU Delft, with extensive expertise on computational imaging mass spectrometry and fusion across molecular imaging modalities.