The Integrated Science Lab invites you to join the conversation at a Lunch Pitch. Anna Johnson has an algorithm that might be useful if you are looking for the most likely sentence, the smallest DNA sequence, or some other least costly structure. Ehsan is finding ways to unify major plant biology databases into a single queryable network. Both are looking for collaborators.
To encourage cross pollination of ideas between researchers from different disciplines, IceLab hosts interdisciplinary research lunches with the vision of allowing ideas to meet and mate. During the Lunch Pitch Season, the creative lunches take place at KBC on a Wednesday. Please note this Lunch Pitch will be in KBF301 (Stora Fokusrummet) instead of Glasburen where it is usually held.
Register to come to the pitch and reserve your lunch by Monday, 23 March at 10am.
IceLab Lunch Pitch registration will open two weeks before the event.
Note! The default lunch option is a vegetarian falafel sandwich. You can choose an alternative lunch (vegan, gluten free, etc) in a separate form that will be emailed to you once you have registered.
Finding the best trees in the data forest
When a structure is generated from a set of weighted rules, a parse tree is created that shows exactly how that structure was generated. The structure can be anything from a natural language sentence to a graph representing a fluid network. It can also be a DNA sequence, or a parse tree itself. For some reason, we might want to learn the least costly structure that can be generated from a certain rule set: the most likely sentence, or the smallest DNA sequence that fulfills the properties enforced by the rules. In that case, we can apply the algorithm Betty that finds the N 'best' (smallest, most likely, least costly, etc.) structures, disregarding any duplicates.
If you think that this in any way can help your research, please reach out, and we can discuss what adjustments to the implementation you might need to make it suit your purpose.
Interested in: Missed application areas, collaborations
Graph-Theoretic Integration of Multi-Omics Data: Enabling Multi-Hop Reasoning in Bioinformatics
Biological research generates vast, heterogeneous datasets — genomics, proteomics, regulatory networks, phenotypes — yet these typically remain siloed in separate databases with incompatible schemas. In this pitch, I present a knowledge graph approach that unifies many major plant biology databases into a single queryable network. The key insight is graph-theoretic: by representing biological entities and their relationships as a typed, attributed graph, we enable multi-hop reasoning — traversing chains of relationships (e.g., gene → protein → interaction → pathway → phenotype) that would be invisible when querying any single source. I'll discuss schema unification at scale, using graph topology to surface non-obvious connections, and how large language models can serve as a natural language interface to make complex graph queries accessible to non-programmers.
Interested in: Collaborations with researchers working on network analysis, knowledge representation, or data integration in any domain. Also keen to connect with people experienced in graph neural networks, ontology design, or anyone reasoning across multiple heterogeneous data sources.
KBC Stora Fokusrummet - KBF301, near the KBC café. Find your way to the venue (mazemap link)
IceLab Lunch Pitches are made possible through funding from KBC for the venue and from Stress Response Modeling at IceLab for their coordination and lunches.
