Genetics Research and Networking

November 26, 2012 8:53 PM

Career mashup: Genetics Research and Networking.

For the last twenty years I've been writing networking software for the various devices (routers, switches, WiFi gear, cell aggregators) that make up the Internet. More recently I've played with cloud switches, but basically I've been doing embedded system networking.

Lately I've been thinking I want to get out of the infrastructure game and work on something related to genetics research. The job doesn't have to be directly involved, I'd just like to feel like I was helping things along in some way beyond providing the internet connection.

I know there are a lot of issues in corralling and sharing the genome data, but other than that I'm flying blind. Could someone with experience in genetics research clue me in where there might be some crossover?

Bonus points: a company with an office in the Silicon Valley.

posted by Tell Me No Lies to Work & Money (5 answers total)

From a strictly IT perspective the big challenges in genomics are computation, data transfer, and storage. Data transfer and storage have simple, throw money at the problem solutions. Computation is a moving target. Analysis approaches change constantly but are rarely optimized. The people designing these approaches need help making the most efficient use of whatever infrastructure they have available. This is often some kind of Linux based HPC cluster running a mix of open source and commercial tools along with a lot of perl/bash/python/etc stringing everything together into pipelines. Be familiar with the tools (bwa, novoalign, samtools, picard, GATK, Galaxy, IGV) , the various sequencing platforms (read up on Illumina HiSeq. There are others, but these are very popular.), the common file formats ( fasta, fastq, bam, bed), the types of sequencing done ( whole genome, exome, rnaseq, mate pair, chipseq, etc) and what they're good for. Read through the forums on SeqAnswers. If you have strong Java, Bash, Perl skills, expertise with a Linux environment, and solid experience with an HPC system ( Open Grid Engine, for instance ) I think you'd have a good shot at finding very interesting work in the field.
posted by roue at 9:22 PM on November 26, 2012 [2 favorites]

Best answer:
I know there are a lot of issues in corralling and sharing the genome data, but other than that I'm flying blind. Could someone with experience in genetics research clue me in where there might be some crossover?

I've recently started working with a researcher who is doing genetics research, although I've been doing IT stuff for a neuroimaging lab for several years now. I'm still learning the pipeline, but the idea is to take samples, segment the data, and then try to match the segments against a baseline genome and find individual variations. The goal is to then match those individual variations with structural data (MRI, etc) and behavioral data.

The software used in this process is fairly rudimentary. In many cases, it may take a variety of steps with many different parameters to get useful results for analysis. I could talk your ear off - but the point is that there is a need in research for good developers who have a grasp of the problem domain. It is my feeling that most of the software is written by people who understand the problem domain better than the software.

I'm sure that there are companies who do genetic analysis for a variety of uses, but I think you might have the most success trying to find work at a research university as research or academic staff in a lab.
posted by Pogo_Fuzzybutt at 9:26 PM on November 26, 2012

First off, keep in mind there's a lot more non-human genetic data than human, with substantially less privacy concerns. A lot of the actual research is done at universities, with big pharma stopping by from time to time looking for interesting chemicals.

I see three things on your resume:
* network protocols and engineering
* software development
* embedded systems

You could conceivably get a job working as a network / system engineer at a bioinformatics cluster, with a specialization in tuning network communication. But it'd be rare to find a place that thinks network is the bottleneck instead of disk or CPU. The networking piece is going to be high perf networking gear (Infiniband, 10 gig ethernet or similar) for internode communications or disk I/O.

To give you an idea of what sorts of biocomputing goes on, there's two major tasks I know of: assembling a genome from parts (shotgun sequencing), and fuzzy search. The software / algorithms I know about these are pretty much designed in a different era, and you wouldn't believe the poor quality of software that bio profs thought was amazing. So software dev experience is a plus, and you could kinda parlay your network engineering exp as a kind of distributed computing platform (ie spanning tree protocol, etc). Because a lot of this needs to transition to something like Hadoop.

But you really need to get some problem domain bona fides, or your resume probably won't make it past whatever HR screening process. Especially within university settings, I wouldn't count on the hiring committee for genetic research to adequately distinguish between what you can provide and someone with a CNA.
posted by pwnguin at 12:15 AM on November 27, 2012

Proteomics!
posted by KokuRyu at 12:15 AM on November 27, 2012

Some great contributions already given. I work for one of the major vendors of sequencing / genomics equipment, and I can say you have a lot of options where you are in S.V.

Heck, I work for Life Technologies, and Ion Torrent is gaining traction, and the newest system (the Ion Proton Sequencer) has a bright future. So you may find something in the Bay Area by looking around.

Illumina has a presence (in Hayward) but from friends have told me (I used to work there) they are not expanding their personnel there, but rather moving them down to San Diego. LifeTech has facilities in Foster City and South SF.

From my perspective there is a high demand for software expertise but of course there are needs across the IT spectrum, and you are correct in your assessment about its promise for the future. (I can't think of anything else I'd truly rather be doing for the next 20 years.) You may want to take a look at this post entitled 'Is there room for 90 providers of genomics software?' on the blog I write up. Many startup companies are diving into genomics, both cloud-based and locally instantiated. Some may become moderately successful, others will be acquired, and most will disappear, but that's what happens.

Feel free to MeMail me if you have further questions.
posted by scooterdog at 9:23 AM on November 27, 2012