LMIB is heading to the Helmholtz AI for Science Conference in Munich!

We are excited to share that our colleagues Anis Ismail (PhD Student) and Saptarshi Chakrabarti (Data Engineer) will be representing the lab and presenting projects from our line of work that they have been helping advance.

Reproducible Multi-omic Benchmarking with Multiverse (June 8, 9am to 11am, Helmholtz Munich Campus)

Anis and Rishi will lead a workshop introducing best practices for benchmarking multi-omic integration methods using Multiverse, our new framework for reproducible benchmarking. The workshop will demonstrate how containerized components enable transparent, reproducible, and extensible computational experiments.

Interpreting Multi-Omic Integration Models with Feature Attribution Methods (June 9, 2:48pm - 3:02pm, Munich House of Communication)

Anis will present part of his PhD research on using feature attribution methods, including DeepLIFTSHAP, to investigate the inner workings of multi-omic integration models. Using MIMA and the prostate cancer dataset from our recent work as a case study, he will showcase a framework for uncovering the biological signals learned by these models and identifying molecular features across modalities that may contribute to prostate cancer biology.

Good luck to Anis and Saptarshi!

If you are attending the conference, feel free to connect with them to discuss their work, exchange ideas, and explore potential collaborations. Stay tuned for the preprints showcasing these works.