The 2025 Society of Biomolecular Imaging and Informatics (SBI²) Conference in Boston (Oct 27-29) brought together the community of scientists, technologists, and innovators pushing the boundaries of phenotypic screening, AI-based analysis, and high-content imaging.
From data to discovery
A shift in focus One message echoed across sessions, posters, and panels: we’re not just capturing more complex biological data; we’re transforming how we extract meaning from it.
The convergence of AI, multimodal profiling, and imaging is redefining what it means to understand cells. From discovering new targets to predicting mechanisms of action, the field is moving fast and SBI² 2025 made it clear that collaboration, openness, and rigor are driving the evolution.
Leadership in action
Nasim Jamali, PhD – President, SBI² 2025 & Director of Morphological Profiling at Ardigen
This year, our own Dr. Nasim Jamali served as Conference President, helping shape the scientific program and guiding dialogue across communities. In her Spotlight Talk, she outlined how morphological profiles derived from high-content imaging can inform bioactivity readouts and mechanism-of-action hypotheses.
Her message was simple but powerful: profiles are signals, not just features. And when interpreted well, they tell stories about cells we couldn’t read before.
Hands-on AI: Translating complexity into practice
The conference opened with workshops and career development panels, setting a tone of inclusion and mentorship.
As part of the education track, Adriana Borowa, PhD, Lead Data Scientist at Ardigen, gave a lecture on AI-based image analysis. She walked participants through:
- What makes an imaging dataset “AI-ready”
- How normalization and alignment affect downstream reproducibility
- How to leverage public models for feature extraction and interpretation
Attendees didn’t just listen, they got hands-on tools, including a ready-to-run pipeline built in Python and PyTorch.
Keynotes that moved the needle
Day 2: Dr. Marinka Žitnik (Harvard University)
Topic: The rise of the “AI Scientist”
Marinka introduced an emerging paradigm: AI agents capable of hypothesis generation, multimodal integration, and interaction with experimental platforms. These aren’t theoretical models, they’re already being used for tasks like:
- Predicting protein-protein interactions
- Patient subgrouping for clinical research
While the risk of hallucination exists, Marinka emphasized that linking models to curated knowledge bases and real-world feedback loops enhances reliability.
Her takeaway? AI is no longer just a tool. It’s becoming a collaborator.
Day 3: Dr. Anne Carpenter (Broad Institute)
Topic: The cell painting journey
Anne traced Cell Painting from its inception to its broad adoption in pharma and academic pipelines. Her keynote highlighted:
- The birth of Recursion and SyzOnc
- The scale of JUMP-CP and OASIS
- New frontiers in rare disease and genotype-to-phenotype mapping (e.g., VISTA, NIH IGVF)
What started as a method is now a movement. Cell Painting has gone from concept to cornerstone.
Emerging trend: Virtual cells and scalable discovery
SBI² 2025 spotlighted the push toward simulating biology at scale:
- Johnny Yu (Tahoe Therapeutics) shared how genome-wide perturbation and single-cell drug response data can inform transcriptome-guided drug discovery.
- Yue Qin (Broad Institute) presented work on hierarchical cell architecture maps, enabling structured interpretation across tissue and disease contexts.
Together, these efforts point toward the virtual cell: a computational proxy for hypothesis testing, drug response prediction, and experimental design.
Poster spotlight: Smarter toxicity profiling
Our team presented early findings on combining Cell Painting profiles with chemical structure data to predict compound toxicity earlier in discovery pipelines. The result: multimodal models that outperform unimodal ones, enabling safer and more efficient compound triaging.
What we’re taking forward
SBI² 2025 showed a field that is:
- Embracing richer biological models
- Demanding more interpretable AI
- Investing in collaborative, open science
At Ardigen, we’re building tools that help researchers move from raw images to biological insight, and from observation to action.
Want to explore phenotypic profiling, AI tools, or imaging data strategies with us?
Let’s talk.
