Computers & Biology: The Early Years
Recently I wrapped up a long career as a computer scientist supporting biological research efforts at a National Laboratory. Working at the intersection of the two fastest-evolving sciences left no time for reflection on the incredible changes that I have been fortunate enough to observe and participate in, but now I can pause and look back a bit.
As a computer science graduate student at a National Laboratory I stumbled into biology by dumb luck: my academic advisor didn't get his operating system computer security project grant renewed (operating systems were written completely in Assembly language back then), so he took a position as the sole programmer supporting the ~150 person biology program.
Minicomputers of that era (1975-77) filled entire 19" racks and were built from discrete logic chips plugged into sockets on circuit boards, with individual wires wrapped around pins on the back side making all the connections. Multi-layer printed circuit boards with etched traces instead of wires were still a ways off, as were integrated circuit chips that contained entire CPUs larger than 8 bits. Since the biologists were too poor to afford a maintenance contract, when something broke a service person would come out whenever they weren't busy with customers who could afford maintenance contracts. When told one time that it would be two weeks for a repair visit, I realized that I had access to a complete set of logic diagrams and a supply of spare chips. So I had an engineer show me how to use an oscilloscope and I figured out on my own how to track down failed chips and replace them. Chip failures were pretty common back then, and I eventually got good enough at repairs that the manufacturer begged me several times to wait to fix the next failure until they could rush a new repair person out so that I could train them. Of course, as is often the case with both computers and biology, by the time you get proficient at something it becomes obsolete rapidly: the next generation of computers had neither socketed chips or wire-wraps, and computer repair was done by "knuckle grinders" who just swapped out complete printed circuit cards.
In early 1976 I got to participate in networking together two mini-computers using a proprietary parallel interface (Ethernet was still being born) and worked with the first color computer graphics used for biology at that Lab. Filling another 19" rack and costing over $50K, this system provide 8-bit color but had zero software. I had to write and debug a device driver (in assembly), write some primitive graphics routines (in Fortran), and then finally code some automated microscope display and image analysis applications. Given the 64KB memory limitations (yes 64 kilobytes!) the image sizes were tiny. A 50MB disk drive (also costing over $50K) was the size of a washing machine, with 16 platters nearly 18" in diameter. When it died a few years later from a head crash over a 3-day weekend, the entire computer room was coated in aluminum dust requiring a hazmat team to clean up.
I returned as a full-time employee in late 1978 after a stint in management consulting in SF following my MS degree. My new boss in the biology program had just come from MIT along with a Unix system on a DEC PDP 11/34 minicomputer. I had read about Unix in grad school and I quickly absorbed programming in C and the Unix shell tools (awk, lex, sed, etc.) Every peripheral we bought needed a device driver to be written, but at least this was accomplished in C instead of assembly. There was email for the first time, using 300-baud phone lines for point-to-point connections with other Unix systems. ARPAnet had a node at the National Lab by then, but its usage was a private club that biology wasn't part of. I also purchased an early Apple IIe computer and interfaced it to a cell counter (I had to write a memo to procurement stating that the personal computer would not be used to play video games, because nobody believed that they might be useful for work.) The device driver I had to write in Pascal remains as the ugliest abuse of a wretched language in my entire career.
By the early 1980s advancing CPU technology resulted in 16-bit microcomputers running Unix that cost about $40K and could have 24 or more terminals connected via RS232 cabling. Using biologists as my "lab rats", I experimented with providing them with terminals and teaching them how to use Unix for command-line data processing with awk and word-processing with vi and troff. By the late 1980s over 50 terminals were in daily use by biologists. It was amazing how well they performed when trained (most biologists, unlike some researchers in other better-funded disciplines, are willing to learn anything to achieve their research goals); but all that computer self-sufficiency vanished quickly when the Apple McIntosh ripened and turned nearly all biologists into Apple zombies, forever doomed to the mercy of applications written by others. (On the other hand, Apple is still thriving today, something which can't be said for Modcomp, DEC, Zilog, Sun or any of the other computer vendors that I used long ago!)
The usage of computers in biology since the 1990s is obviously well-known since so many from that era are still in the field, but few scientists still active today were around when computers in biology first replaced the hand calculators that were used for all data analysis. Although I was amazingly lucky to work with great teams on the Human Genome Program and numerous biodefense programs, it is fun to look back at the primitive early days of computing in biology and marvel at how resourceful we had to be to accomplish anything useful in a world without an Internet or Google.
(Tom Slezak was a student employee at Lawrence Livermore National Lab from 1975-77 and a full-time employee from late 1978 until September 2018. He is now the CEO and Co-founder of KPATH Scientific, LLC, providing technical and managerial consulting to a number of biotech startups.)