Blog Post
April 14, 2018

2018 Kidney Cancer Hackathon: A Patient Perspective

(in response to a question from Tony LiVigni here)

RareKidneyCancer.org’s (free) May 18, 2018 p1RCC hackathon is hosting teams consisting of Biologists and Data Scientists.  We will also have an NIH TCGA representative attending.  TCGA (The Cancer Genome Atlas) is a genomic data collection of 33 cancers from 11,102 tumors including 291 papillary kidney cancer tumors.  TCGA has been the subject of many pan-cancer studies.  We also have an an NCBI coach (video) whose goal is to help the teams produce software that provides more insights using TCGA and my blood and my p1RCC genome/transcriptome as a reference.

I have been working very hard to encourage a pan-cancer approach from the invited groups.  My hope is that by studying the relationships between ALL cancers, we can get some insights into the rarer ones (including ours).  If this idea works, it may help overcome the pRCC funding deficit.  Another thing that would increase research is more patient data.  As I argue in http://rarekidneycancer.org/blog/kidney-cancer-month-value-public-data, since most patient data is kept private, data that is made public (like mine) receives more attention from academics.  For patients that have already been sequenced, this may only require signing a consent form.  Contact me at bill@rarekidneycancer.org if you have questions.

So what kind of work can we expect to see from these pan-cancer studies, emphasizing pRCC?  Clemson’s Alex Feltus’ work (based on t-SNE clustering) will (hopefully) relate drugs used in treating other cancers to drug candidates for pRCC.  Stanford’s Avantika Lal’s work (described here) relates TCGA data to patient survival.  E.g. I can run the code to see the average survival of people who have a papillary kidney cancer subtype like mine.  SLAC’s (QuantumInsight.io’s) Marvin Weinstein’s work (based on DQC clustering) has discovered similar relationships between stage and patient age.  By bringing these teams together, I hope to cross check them and see if the results of one can better inform the clustering techniques of the other.

 

Add new comment