November 17, 2017  |  Announcements, CITRUS, Conferences  |  By  |  0 Comments

On the Importance of Biomarkers and Integrated Data – Lessons from the Society for Immunotherapy in Cancer Meeting

Dr. Kat Drake, 2017-CircleLogo-shadow-light-v2our Director of Informatics just got back from the annual meeting of the Society for Immunotherapy of Cancer (SITC). Read more for her observations, notes and key takeaways from the world’s foremost conference on Cancer Immunotherapy.

SITC 2017 was a great conference with many interesting and packed sessions.

Machine Learning Tools are Great for Biomarker Discovery

Biomarkers were, of course, a major theme throughout the conference and it was nice to see several great examples using Cytobank for biomarker discovery and analysis in clinical contexts. My favorites included:

Poster #130: In situ vaccination with Flt3L, radiation, and poly-ICLC induces a potent immune response in patients with follicular lymphoma
Poster #130: “In situ vaccination with Flt3L, radiation, and poly-ICLC induces a potent immune response in patients with follicular lymphoma” by Thomas Marron of The Icahn School of Medicine at Mount Sinai.

Duhen Poster 146
Poster #146: “Co-expression of CD39 and CD103 identifies tumor-reactive CD8 TIL in human solid tumors” by Thomas Duhen of AgonOx

Data Integration is Complex but Critical

Illustrating another major theme in the sessions I attended, Thomas Duhen’s poster integrated several data types to characterize and understand the biological mechanism behind differences in clinical outcomes. Tom Gajewski from the University of Chicago gave one of my favorite talks on this theme, “Integrating Multiple Dimensions of Genomic Data as Immunotherapy Predictive Biomarkers.” He and others throughout the conference showed examples of integrating multiple data types to better understand anti-cancer immunity, each using multiple data types to put together a story. Tom warned that it’s important to keep in mind that the interactions between the different variables we measure can be complex and difficult to identify when you look at each variable independently.

At Cytobank, we’ve been working on using our machine learning tools to help identify and quickly understand these interactions in complex and integrated data. For example, our poster with Nanostring was a big hit, demonstrating the use of DROP and viSNE on RNA and protein data from sorted cell populations.


“Kit & Kat”: Kit Fuhrman of NanoString and Kat Drake of Cytobank Present Poster #40: “Deep proteomic and transcriptomic analysis of sorted T cells with a simple, integrated workflow.”

I’m excited to see how our work and that of our customers will use these tools to contribute to efforts to integrate and analyze multiple data types going forward!

Learn more about how you can: