Blog Post
March 11, 2020

2020 Kidney Cancer Hackathon: Results

Summary's key mission is to enable creation of treatments for rare kidney cancer patients.  Unfortunately, most cancer treatments are "population-based", which assumes that a disease is (mostly) the same among all patients labeled with the same disease type (e.g. p1RCC).  But often, each individual tumor has its own set of unique mutations, and this variation likely makes similar types of cancer respond differently to the same treatment.  Not only tumors between patients, but within the same patient as time progresses.  This is because each patient's body is its own ecosystem, and as the ecosystem changes (for example via the addition of various drugs) parts of the cancer die, but minute, genetically-distinct non-targeted parts of the cancer do not, and so "select out" to thrive in the new environment.

So how do we deal with this?  Currently, assumes that although the treatment may not be the same between patients, the plan for arriving at a treatment is.  We outlined the current version of our plan in a prior post. In brief, well before the hackathon, we gathered the data, verified it and gave it to the research teams.  The teams explored the data and brought their findings to the hackathon.  At the hackathon, they then reviewed the research results, identified potential targets and mapped the targets to potential therapeutics.  


The Tri-con hackaton spanned 3 days and was the first hackathon run in conjunction with a commercial tradeshow.  This produced several notable differences from prior hackathons, but the major positive difference was that Dr. Jeannette Koschmann from Genexplain left the show floor and joined the Clemson team (Reed Bender, Ben Shealy and Benafsh Hussain from Dr. Alex Feltus' group) shortly after the hackathon started.  So the Clemson team was able to take the UCSF/Yale data and produce a set of candidate genes which (hopefully) account for the disease.  Then Dr. Koschmann used the GeneXplain software to map from this gene set to some suggested therapeutics.  We were unable to do this in prior hackathons, and were only able to do it this time due to the tricon format, which enabled the kind participation of Dr. Koschmann and GeneXplain.  

As such, we were able to hit our primary goal, which was to make it all the way to the plan's activity 9 "discuss potential therapeutics".  Clemson is generating a paper from the research and discovered a new potential collaborator in GeneXplain.  GeneXplain was able to demonstrate how their technology worked, has some progress to discuss with specialists in the field and TRI-con/RTTP can rightfully claim credit for enabling it all.  Those interested in more detail can examine the documents generated below at the hackathon.

Future Work

Going forward, we hope to combine the advantages of both TRI-con and prior hackathons.  E.g.  Recruit (more) teams via direct email as well as from the tradeshow, allow remote team participation and make cloud storage available well before the show.  Oh, and one final "learning", next time, we'll try not to hold a hackathon during a pandemic.

Though we got further along in the process, we still feel that multiple teams are necessary since that structure introduces both co-operation and competition between multiple un-correlated methods; E.g. The 2018 hackathon had 17 different teams working separately at each step and comparing results at the end.  The 2018 effort produced 17 different targets. As such, we now need to cross check the one result from this hackathon to see how good it is.  This motivated us to add activity 10 to the original plan ("verify target and therapeutics").  Our hope then is that the plan, coupled with a hackathon, will work like a distributed tumor board, a group of doctors and other health care providers with different specialties that meets regularly to discuss cancer cases and share knowledge. In the best case, the board determines the best possible cancer treatment and care plan for an individual patient.  Once we've run through this a few times, the goal will be to automate as much of it as possible.
Finally, as the Acknowledgements indicate, there are a lot of kind individuals who have devoted time and energy to move the ball forward for just this one case. 
In addition, Pete Kane is in the process of creating a permanent home for the co-ordination effort for other patients and Dr. Alex Feltus is integrating patient centered data into his research and educational materials. In his research lab, he is integrating individual patient data with open source patient data from projects including The National Cancer Institute, Genotype-Tissue Expression, GnomAD, and SPARK projects to isolate the biology and treatment options for the individual. On the educational side, he is integrating these datasets into experiential virtual labs in the undergraduate and graduate courses he teaches at Clemson as well as the broader planet using the Praxis AI platform His goal is to have people learn how to do data-intensive cloud computing on real datasets, making discoveries that impact lives, and unleashing the power of the new biology.


Thanks to.

The TRI-con 2020 Team: Bean Shealy, Reed Bender, Benafsh Hussain (who was into the mask thing waaaaay before it was cool.), Dr. Jeannette Koschmann

202003 TRI-con Team

Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.