I discuss the general value of making your tumor and blood data public here. But what am I doing with the public data I can access?
I’m trying to apply the “network effect” to all cancer research.
Anyone who has used social media knows about the network effect. The classic example is the telephone, where, as more users join the network, the value of the network increases for each member. So how do we make that work for cancer research? The basic idea is simple. Enable researchers to automatically share and run interoperable, public domain software experiments on free, standard, patient-contributed public data. Now, this only works for software experiments, and even then, not every public domain experiment interoperates; however, a lot of experiments do.
In this network, participating patients get continuous attention since their data can be run against new interoperable experiments as they are added. In fact, if a lot of rare disease patients “step up” and donate their data to the network’s data pool (which is completely allowable under HIPAA and stored on openhumans.org), they should receive an outsized return, since researchers will likely concentrate their research on the free public data, not data “siloed” behind paywalls. For example, interoperable code written to find therapeutic targets for colon cancer will be rerun to find therapeutic targets for papillary kidney cancer as soon as the papillary kidney cancer data is contributed.
In this network, participating researchers have an opportunity to publish a limitless series of papers with collaborators inside and outside of their specialty since their contributed code can be re-run whenever new data is entered. In the above example, once the researcher contributes code (on github) to the network to find therapeutic targets for colon cancer, it can be automatically run on cancer data submitted previously or in the future. If something of interest is discovered, the result is flagged and the researcher is notified, as are patients and researchers who registered interest in the disease. This enables them to get academic credit, collaborators and grist for research papers if they so choose.
Nonprofit organizations like NCBI are helping provide initial content for this network via a series of “hackathon” gatherings where patients with particular diseases spend a weekend sharing their genetic data with a lot of biology researchers and “big data” specialists who run these experiments.
But once you have a database of cancers in place, these interoperable experiments do not have to examine just “one cancer at a time”. One project in the upcoming hackathon is doing a new type of clustering of RNA-sequencing data of 60,482 genes measured in 11,102 tumors across 33 cancer subtypes (including 3 kidney cancers) from TCGA (The Cancer Genome Atlas). Why? Because prior experiments of the method revealed previously unknown “bridges” connecting one subtype to another. This may help determine whether drugs targeted for one cancer will work with another.
The GEM (“Gene Expression Matrix”) input to this clustering algorithm was created by Clemson’s Leland Dunwoodie, and he is making it freely available. In fact, many others have contributed to elements of this effort -- free of charge, including: Leland Dunwoodie, Dr. Feltus' group at Clemson, QuantumInsights.io, WUSTL's James Hsieh, UCSF's Max Meng, the physicians of RareKidneyCancer.org, sv.ai, openhumans.org, TCGA, the NCBI, and Google.
Imagine. All cancer patients, standing shoulder to shoulder, contributing public domain data to a common pool. Data starved researchers exploring that pool and contributing public domain code to it that anyone can run. A Social Network used for Social good bringing them all together. Wouldn’t that be something to see.
If you are a patient interested in contributing your data to this effort, or if you are a researcher interested in contributing code or getting Leland’s GEM, if you want to join the next hackathon or hold one for your own disease, or even if you simply want to keep abreast of how this develops, please let me know at email@example.com.