Getting started in the Bioinformatics industry.

1. Growth projected

2. Getting Started

I am seriously considering volunteering time and talent to a university in the hope of making industry contacts. For example, Raleigh, NC's "The News and Observer" ran an ad for a bioinformatics programmer at Duke University. Even though I might not qualify for the tight credentials required for that position, maybe if I call up the professor and ask to volunteer my time, he might be willing to take me up on my offer. I could do this for a university, but would not consider doing this for a "for profit" institution, unless maybe I was a recent graduate having a hard time gaining employment. Personally, I have come to believe that if one offers free labor, then he does not value his time and efforts enough, and is doomed from that point on. We are living in a competitive capitalistic society, and hence, cannot afford to undercut our going rate, else we end up without our science and our livelihood. Also, there is an element of negotiation here, which I cannot emphasize enough. One has to spend his time marketing his skills and ability, and find someone in a position to pay for those skills. If you are working free, that means that you are not selling yourself enough. I have seem software engineers turned into marketeers, so if that is our destine, than I want to at least be paid for it.

Perhaps this is easier for me, since I have an eighteen-year background in software development (C++), graphics & CAD/CAM, but think that other software engineers can do it as well. I have made only one career transition in my life, from Math to CS, and looking to get noticed. For example, if I manage to rework some things I did on another project, I am hoping to get a paper published in "IEEE/ACM Transactions on Computational Biology and Bioinformatics", ISSN 1545-5963. However, I have held a dozen jobs in a variety of industries, so I look at this as yet another transition.

My first attempt along these lines was to look at the work "Linus" (named for [1970's Vitamin C guru], 2 nobel prize winner [concept of chemical electronegativity] Pauling) being worked on by Prof. George Rose at Johns Hopkins University. The Linus project is an attempt to predict and model protein folding. My fundamental motivation for starting the Linus project was based on reading an article in Science magazine, and thinking just how badly engineered most of the software that I had seen had become, I figured that software reengineering a good scientific idea would improve it considerably. Unfortunately, Linus was not as successful as advertised (Science, Dec 1999), and as such has fallen into the category of less than as successfully as anticipated. Well, those are the breaks, especially in R&D.

Another shareware project recently discussed is Insight ( http://www.itk.org, http://caddlab.rad.unc.edu/ & http://sourceforge.net/projects/dspace/), which is a collection of computer visualization code for the Visible Human Project. Febuary 2005, Vol 48, No. 2 issue of "Communcications of the ACM", pp.55-59 has an interesting article on the software package.

Open source software has caught mainstream attention, if for no other reason than closed source software is expensive and faulty too! Microsoft may conspire that Linux will just disappear, but I doubt that will happen. Often, bioinformatics companies look at the successful products developed by individuals and univerisities, and decide that there is money to be made by re-packaging ideas and concepts in fancy wrappings (GUIs, languages, and OSs). If I were to make a business in this manner, I can look at offerings in commercial software and add similar capability to open source software, and sell one's customization services. For example, one commercial developer commented that the biggest challenge to growing its bussiness is that competition comes from the open-source communitity. The target audience for the commercial software is pharmaceutical researchers who come from an academic background, and thus have exposure to good sciencs and can identify accurate results. However, most university scholars cannot affort commercial products and work with previously developed open source software. Hence, once a researcher lands a position at a pharmaceutical company, he is not going to be too interested in learning a commercial product (especially with project deadlines of his own), and prefers to use the [open-source] ones which he already knows. This appraoch of repackageing ideas is true across many industries. RedHat does not write open source software, it just packages tested components together and ask a modest distribution fee for its services. I believe there is the business model to be copied!

Most pharmaceuticals and bioinformatics companies will want university credentials, and have no problem attracting talent from well-established institutes. One could take bioinformatics classes at Wake Forest University, but in my opinion, the credentials which will count for considerably more are those taken at recognized top rated universities, e.g. Duke, UNC, or NCSU. If you are working, there is a good chance that your employer will pick up your tuition for classes, at least at state universities (having cheaper tuition than private universities). Unfortunately, if you are not working, them you have 2 options, taking some classes at a cheaper institute and paying out of your own pocket, or going to MyU, my pun on individualized education by independent research. I believe that some people (but not headhunters or HR) will take this seriously if you can demonstrate a working knowledge of the topics and research involved. For example, in graduate school, most of the topics I studied were independent in nature, and even if a topic was a part of the curriculum, a student had to review several sources in order to be to understand the material. That, and the fact that most classes had projects outside of the immediate syllabus. When you are going to be interviewing for a position, the interviewer will want to see that you have read more than Bioinformatics for Dummies! It would be very helpful if you could research a topic of interest, and try to get an article published in a trade magazine, which does not have the same high expectations as a professional referred journal. At times, it is possible to look at work that you did in one field, and apply that to another field. For example, Rivest, Shamir, and Adleman (RSA) developed encoding algorithms for communications. Later, I have see Dr. Brudno apply the inverse RSA algorithm to sequence alignment algorithms.

Some authors explain some topics to some people better than others do. For example, there are a great number of books published on the computer language C. The choice of books spans from Sam's Teach Yourself C in 24 Days (weighing more than a NYC phonebook) to Kernigan and Ritchie's The C Programming Language in less than 78 pages. However K&R is written by geniuses for geniuses. It is a reference book where you want to get to the point very quickly and succinctly.

Secondly, there is a lot of grunt work in industry. I worked at IBM Watson Research. This was a mixed blessing since I did not have a Ph. D., I was given the less than glamour tasks, e.g. reorganizing code, modifying makefiles, debugging and testing. The good thing is that I had a job. However that IBM job was less than all that I hoped for because I had previously worked at a small start-up in which I was the one who was doing the interesting algorithm development (CAD kernel development). Unfortunately, the start-up went bankrupt, and I needed another job! IBM was a good experience in learning things from other people, but I wished I could had have been the top contributor.

3. Technologies, Trends, and People

In order to keep current on the industry, one has to keep up on the literature, trends, and people making headway in the field. Opportunities will dry up and money will be diverted to other endeavors, so keeping up is a necessity. One has to visit the library and read what is happening in his field and those around it. Try to attend one conference a year or a local user group. Try visiting university talks. Yes, most of the time, nothing will come out of it, but then, if you don't try, no one will really try to promote you but you.

4. Human Social Skills

Even though we are looking for jobs in the high-tech, we must always be aware that our appearance counts for as much as our brains and talent. I have worked with several software engineers who were very smart, but really neglected their outward appearance. I understand wanting to dress comfortably, but sometimes, some people take it to an extreme. I have known software engineers who would not bath every day. Yes, he might be smart, but I doubt that too many people will want to work with him. Unfortunately, all technologies go stale, so even if you are a top performer in your field, there is a good chance that what you do will be obsolete in a couple of years. I have meet a great many a people whose topics and techniques were dated, but who managed to work the social circles and sell themselves to top management as if they were a rare commodity. It is always important to communicate effectively and to be able to write and make presentations. You might have a great idea, but if you cannot write a grant, you are not of much use to the movers and shakers. An often recommended book is Dale Carnegie's, "How to Win Friends and Influence People". I don't like the "kiss up" attitude that Carnegie stresses, but it does sometime work, but most often backfires because people take you as too much of a pushover. There is a lot of work in documentation, presentation and testing. Software in usually enormously complicated, and often counter-intuitive by design or functionality. Unix was vastly superior to VAX/VMS in being a small robust OS (in just a four man year project), but Unix was often inconsistent and counterintuitive. Most bioinformatics projects are developed by Ph. Ds in chemistry and biology, who often don't have the time and experience in developing software. Face it, a Ph. D. is a grueling experience with rewards, but the aim is not to write clean software. If someone was willing to pay me for my time, I would like to document and test everything that I do, but often times there is a great deal of work to be done to get the software working in the first place. I am sure that most software engineers have been faced with this obstacle at one point in their career, and sometimes there is hope in selling one's ability to document, test, and market other people's work. At IBM, I was often asked to make presentations to user's groups about software which was written a while ago, but which no one had the time to demonstrate. Also, do not underestimate the publicity value of writing articles. Not that it pays much, but consider writing about a topic about which you know a lot, which might have appeal elsewhere. There is free publicity, and will be far more effective than cold calling and mailing out marketing brochures (not that you should not do that as well).

5. Stories from the Future

Last year, while working at IBM, I went to a talk on applications of [BlueGene] supercomputers to biological problems, given by Dr. Jerome Rice, [one of my interviewers at Physiome] on modeling of the human heart. On Wednesday, Feb 23, 2005, Prof. Raimond L. Winslow from JHU came and gave a talk about his modeling of the human heart, which was similar to that given by Dr. Rice. As the talk progressed, I realized that I had read a paper by Prof. Raimond L. Winslow when I had interviewed [and received a job offer] at his startup company, Physiome Sciences in Princeton, NJ in 2000. Physiome was a company that modeled the effects of drugs on the human heart based on Prof. Winslow and Nobel's work. Unfortunately, Physiome promised too much and charged too much, and ended up going bankrupt about 2 years ago. After the talk at Duke, I spoke for an hour with Dr. Winslow about Physiome and biological modeling in general. In retrospect, the Physiome idea was too elaborate and the goals too ambitious for today's technology. However, Physiome raised $50M in VC money, and turned down another $50M because it did not want to give away too much ownership. $100M is a lot of cash to play with, so I wonder if the problem was not on spending too much money on computers and not enough on the problem at hand. In 2003, Physiome Sciences merges with Predix.
One of my good friends, Dr. William Luken, a quantum chemist from Yale (now at IBM) who has made a career change from bioinformatics (before there was such a field, as he modeled electon distribution regions when one electron's position was fixed) to computer programming (graphics, sounds, and multimedia). Bill once told me a joke. A guy lost his car keys and was looking for them at a street corner. Another guy was walking by, and asks him where did he loose the keys. The first guy responds, back there in the alley. So why are you looking for them on the corner, asks the second guy. Well, the street lamp is here on the corner. I hear similar stories about using molecular dynamics algorithms to solve protein-folding problems. Molecular dynamics algorithms use ball, spring, & mass systems to model DNA chains. The better approach is to model the system with Schrondinger's Wave Equation, but stochastic differential equations for quantum events are beyond anyone's current mathematical attempts. 

Another of my good college friends has a B. Sc. degree in physiology but got smart and went for a JD law degree and now is working on patents and trade regulations.  He jokes that it is perhaps more lucrative to be involved in patents and regulations than to be discovering patents and regulations.

While working at IBM, I also had a chance to meet Prof. Richard Karp at Washington University in Seattle, WA [Ph. D in Applied Mathematics from Harvard], who made the career change from math to biology in the 70's.  This is all right at a university research setting, but talking to my wife (who has an Ph. D. in Biology and is an MD) about bioinformatics, she thinks that it seems as if computational biologists are very far away from real biology.  I showed her my class syllabus. She remarked that that was all mathematics and only tangentially related to biology.