User:WillWare/Whole-cell simulation

Whole-cell simulation is the simulation of a complete biological cell, ideally a human cell. This could be very good for medical research and science education, and could help advance the goals of the SENS Foundation, which wants to help us all to live for hundreds of years. I'm on board with that.

Mixed-mode simulation
I want to use mixed-mode simulation to try to simulate as much biological stuff as possible. My thought is to use GPUs and CUDA to accelerate the computation.

Also see Physics engines.

Could this be my day job?

 * Cygnus subscription model
 * Periodic releases are available for free download
 * Maybe the source code release lags the binary release by several months?
 * Subscribers pay for consulting, maintenance, customization
 * Encourage subscribers to share code with one another, but not the general public
 * Corporate sponsorships:
 * Find corporations that want good public relations from supporting education and research.
 * GPU manufacturers? Big pharmas?
 * If connected to the web, the software pops up a small sponsor logo.
 * Do lots of demo videos that somehow include sponsor logos.
 * When appearing in public, wear sponsor's swag.
 * GPU consultancy
 * Offer training
 * Set up and/or certify hardware
 * Write code for people
 * Others are doing this: 1 2 3
 * Can't go directly to computational biology consulting
 * I don't have the academic credentials
 * GPU consultancy can fund the compbio work
 * If this doesn't work out as a source of income, what other good can come from it?
 * It could help me get a more interesting job.
 * Maybe get a job with NVIDIA
 * Familiarity with GPUs and GPU clusters might extend my career.

Gaining scientific credibility
Assume I start out doing vanilla GPU consultancy for people wanting supercomputers, and the computational biology is a background activity that doesn't generate income, at least initially. It will then be necessary to gain scientific credibility for that work. Domain experts will initially be mistrustful of a mixed-mode simulation. So you need two things.
 * Objective tests that demonstrate the approach does not diminish accuracy, in the region where high-res simulation is done.
 * Lots of input from computational biologists and chemists, lots of transparency.

You need input from biologists and chemists to figure out what are the relevant mechanical properties and phenomena that need to be modeled. Start with mechanics and electrostatics and van der Waals, and then ask them if there are any missing phenomena.

I might need to talk to a conference organizer and book a hotel for a weekend. But there would be lots of preparation to do before that, so that everybody is on the same page about what it's all about, and what contribution they can offer.

Related pages

 * GPUs and CUDA
 * Physics engines
 * Mixed-mode molecular simulation
 * Whole-cell simulation