Cancer immunogenomics

Eduard Porta

Team
Publications
Current Grants

OVERVIEW

The Cancer Immunogenomics lab studies cancer as a dynamic, spatially organized system. We combine experimental spatial omics with computational modeling and explainable AI to understand how tissue architecture shapes tumor behavior, immune interactions, and therapy response.

A central premise of our work is that many clinically relevant cancer phenotypes are emergent ecosystem states: cells with similar molecular profiles can behave very differently depending on where they are in the tissue, which cells surround them, and how those neighborhoods reorganize over time and under therapy. Our lab develops both the experimental methods to measure these states and the computational frameworks to interpret and manipulate them.

COMPUTATIONAL LAB

Spatial Modeling, AI, and Translational Interpretation

Our computational-lab effort develops analytical methods and conceptual frameworks that convert spatial data into biological and clinical insight. We integrate spatial transcriptomics and proteomics with histology, genomics, and clinical annotations to uncover how tumor ecosystems are organized, how they change under therapy, and which spatial features predict outcome.

RESEARCH DIRECTIONS

Explainable AI - We build interpretable machine-learning models that connect tissue architecture to molecular states and therapeutic response, focusing on features such as neighborhoods, compartmentalization, and proximity-based interactions.

Spatiotemporal Hallmark Ecosystems - We develop frameworks that reinterpret cancer hallmarks as spatially organized and dynamic programs, enabling ecosystem-level descriptions of tumor state and transitions over time and treatment.

Spatial Engineering - We develop theoretical and computational models that frame therapy as an intervention that reshapes pathological tissue architectures—moving beyond static atlases toward predictive and design-oriented spatial oncology.

Multimodal integration - We build unified models that jointly leverage spatial omics, histopathology, proteomics, and genomics, allowing us to link molecular alterations to functional, spatially manifest ecosystem states.

Beyond spatial - The lab also contributes to broader computational cancer biology, including large-scale proteogenomic integration efforts (e.g., CPTAC) and privacy-preserving AI approaches for clinical prediction (e.g., federated learning and homomorphic encryption).

WET LAB

Spatial Omics in Tissue Context

Our wet-lab effort generates high-quality spatial datasets across human and mouse tissues, with an emphasis on reproducibility, tissue quality, and experimental design. We use spatial transcriptomics and multiomic approaches to profile gene and protein expression while preserving tissue architecture, enabling analyses at spot, cellular, and subcellular resolution.

Sample Quality Control
All projects begin with rigorous assessment of tissue suitability:

RNA extraction and integrity assessment, including DV200 or RIN measurements
Histological evaluation using H&E staining to assess preservation and guide region selection

Technologies

Visium family (whole-transcriptome spatial profiling)
We use the Visium ecosystem for discovery-scale spatial transcriptomics while preserving tissue morphology:

Visium CytAssist – Whole-transcriptome profiling of FFPE and fresh-frozen samples at spot-level resolution
Visium HD (probe-based, human & mouse) – Whole-transcriptome profiling of FFPE, fresh-frozen, and fixed-frozen tissues at single-cell resolution
Visium HD 3′ (capture-based, all species) – Whole-transcriptome profiling of fresh-frozen tissues using 3′ poly(A) capture and reverse transcription

Xenium (high-resolution in situ profiling)
For cellular and subcellular mapping directly in intact tissue, we use Xenium in situ (FFPE and fresh-frozen compatible):

Xenium V1 – Targeted panels up to 480 genes
Xenium Prime 5K – Broad discovery panels up to 5000 genes in human and mouse
Xenium Multiomic – Combined gene + protein detection (up to 480 genes and 27 proteins)

Equipment