CV Posted

My CV is now available for download from Google Docs.  If you'd like to hire me as a consultant to help you with your code, algorithmic, data or bioinformatic problems, please get in touch!

The Data Afterlife

I was thinking a bit about the picture I used on my blog post yesterday.  If you don't know what it's of, that's where you go when you die.

I'm not joking.  That was in the morgue of an abandoned hospital.  There were dozens of colossal tissue storage cabinets, each with dozens of drawers, and each drawer again with dozens of tissue samples, one for everyone who ever died at the hospital.  When the hospital closed altogether, this collection of tissue samples--thousands of them in total--was simply left to rot in the morgue of the hospital.  In spite of this time consuming data collection effort, these slides were probably never examined again.  Nothing was done with the data, yet each little slide was someone's life.

I don't believe in an afterlife, and I don't believe in ghosts, but somehow looking at those drawers of slides was a solemn experience, like witnessing an injustice long after the fact.  These people gave of themselves ostensibly hoping that this could be a dying gift to science and humanity, and instead their sample was unceremoniously abandoned in a hospital basement.

Maybe that's a bit extreme.  I feel no emotional attachment to my skin flakes in life or death, but then not all of my skin flakes are meticulously tagged, numbered and stored in a tissue storage cabinet. That said, I had the same solemn feeling from finding those cabinets again lately.  I've been examining data from The Cancer Genome Atlas and realised that each of these data points was someone's life.  Hundreds of factors--clinical data, time until death, even genetic information--all made up someone's life, and they nobly allowed me to have the data so that others might not suffer their fate.  In the case of The Cancer Genome Atlas, their data is not rotting in a basement somewhere and never will be:  anonymous data is placed in a public location where all researchers everywhere have the opportunity to use it to make the world a better place.  That's the way it should be.

Many of my colleagues at Sage have a hard time getting other researchers to share their data.  For the researcher who hordes data, that same data may represent their next paper, their next grant, or their job security.  That data is also a precious gift granted by a patient who cared about none of those things, and really just wanted to do their part to advance science and save lives.

For those who do not make their data easily available to all, or sit on data for years, I hope you'll keep this in mind and do your part to make your data open-access whenever it is ethical to. 

The Maw of the Beast

I come from a Physics background, and before coming to Sage had never given any thought to the problem of cancer treatment before.  Looking at the marvel that is the human organism, it was previously my assumption that no cell could survive massive insertions, deletions or substitutions in its code.  In fact, I assumed that the difference between cancer cells and normal cells was a single mutation in a promoter region or in some signalling protein.  Certainly, based on the degree of genetic similarity we see in humans, the cells in this complicated organism that slowly evolved over millions of years couldn't survive a complete reorganisation of its genome... right?

So wrong.

I was reading "The Biology of Cancer" by R. A. Weinberg, and looking at Figure 1.11 is looking into the maw of the beast.  They had fluorescent probes such that each of the human chromosomes would show up as a different color in a normal cell, and then applied the probes to a cancer cell.  The original genome was so jumbled up in the cancer cell that it wasn't even clear what chromosome each piece might have been in the first place!

In fact, cancer cells evolve just like their very own organism.  While the original defect might be a point mutation, the tumor evolves through several phases before finally becoming a rapidly-spreading malignancy.  The evolution diverges so dramatically that part of a tumor may respond to a drug and part may not!  To really drive home the point of how different the cancer cells have become versus normal cells, cancer cells may not even metabolise food in the same way that normal cells do! 


So, that brings me to why there's no cure yet.

• The cancer is constantly evolving to evade whatever assault from the immune system or drugs you apply to it.  Because of its enhanced rate of growth and its loss of DNA repair mechanisms, two hallmarks of cancer, it can evolve resistance to treatments as fast as any pathogen.
• Your normal cells have evolved a complex web of regulatory interactions over the millions of years that led to your existence.  Even if we had the life-saving treatment for a cancer on the shelf among all of the available chemotherapy treatments, we may have a very hard time determining which one to give the patient because of the number of things that are available to go wrong with cells!
• The tumor evolved in a unique environment:  your body!  In its evolution, the cancer may have learned how to exploit the correctly-functioning mechanisms of nearby tissues in order to further its proliferation.  So, if the normal cells are part of what's going wrong, then breaking the cycle of malignancy may necessarily involve toxicity to your normal tissues!

These are challenges that I had never considered before coming to Sage, but they're also things that we're working on.  Cheap gene sequencing may be able to give us access to biomarkers that were previously invisible.  Every month more papers on the body's regulatory networks are released and better analysis methods are developed.  Further, amid these myriad things that can go wrong, as a Physicist who loves generality, I can cling to one absolute amid a phylogeny of complex modes of failure:

• Cancer is always a regulatory failure.

By understanding gene regulation, we can understand cancer.  By understanding it, we can start bringing better treatments to patients.

The GOTO Guy

You can follow a project I'm working on which will bring the versatility of HDF5 on my GitHub account, theGOTOguy, or on Sage's GitHub account.

SVN has been around for a long time, but Git is the future of collaborative software development.  In the past, I've set up SVN repositories for use on collaborative projects and ultimately found that they went unused.  I've felt that it's harder than it should be to create a repository, fork and branch in SVN.    Maybe there's a reason for this--ostensibly, most of your coding should be happening on the core project.  That said, the "social" tools that make communicating on a project, pushing and pulling fluid and the rewarding statistics on Git make collaboration and experimentation easy and even fun.  If you use SVN and haven't tried Git yet, I strongly recommend giving it a shot.

Sage's Synapse project will be to bioscience what Git is to coding.  By making code and data available, accessible and social, we will democratise research and facilitate more and better science, collaboration, and bring results to the clinic faster.