Where does all the computer power go?

Source: "Tracking HIV Evolution," Regina McEnery, IAVIReport, Vol. 14 (3), May-June 2010

The HIV retro-virus is notoriously variable. There are as many as 100,000 variants of the HIV virus in a single infected individual. There are data on the viruses found in at least 250,000 people. It is important to understand the viral population infecting specific human populations as scientists seek to find HIV vaccines. Moreover, the evolutionary history of HIV provides an important input to understanding the source and evolution of the epidemic.

Super computers have been used for some time to understand the population genetics of HIV. I quote from the cited article:

 Last year, through a unique arrangement that allowed a handful of scientists access to LANL’s latest supercomputer, the Roadrunner (pictured below), before it was moved to a classified computing network, Korber, computer scientist Marcus Daniels, and physicist Tanmoy Bhattacharya compared the evolutionary history of more than 10,000 genetic sequences from more than 400 HIV-infected individuals to try and identify common features of the virus that is transmitted and establishes infection. This work was done in collaboration with the Center for HIV/AIDS Vaccine Immunology (CHAVI), of which Korber is an investigator. CHAVI collected the samples from both acutely and chronically HIV-infected individuals from around the world. The samples were used to construct the world’s largest phylogenetic tree, with the end goal of identifying similarities in HIV sequences from samples taken during acute and chronic infection. A single HIV-infected person can have 100,000 different variants of the virus circulating throughout their body, so understanding how these variants branch off from the initially transmitted virus is important for the development of vaccine candidates. To build such a tree, LANL researchers needed Roadrunner‘s processing capability. Roadrunner does 1.042 petaflops, or a quadrillion calculations per second, using 122,400 processors. To gauge the power of Roadrunner, consider this: It took a single week to run a calculation on Roadrunner that the fastest supercomputer a decade ago needed 20 years to complete. 

The computer shown above is a multi-million dollar device. How many developing nations have that kind of computing power? How many of them could or would devote that kind of power to understanding the nature of the HIV virus? This is an indication of a digital divide!

Incidentally, this is one of a series of postings titled "Where does all the computer power go?" You should be able to track the  series using the search function, inputting the title.