jahayes@miavx1.acs.muohio.edu (03/17/91)
It seems to me that this discussion is wandering a bit far afield from the charter of this group. Let me try to bring it back (though I admire the thoughtful character of the very long post earlier this week; I extracted it and posted it to our local computer discussion group to try to stimulate some discussion more useful than the usual junk...) to something more biological. The question, as I see it, is how computer platforms serve the biological research community. In my department here, where I pass for the local computer whiz (hee, hee), there is (obviously!) a real lack of computer smarts. And yet, these people need to use computers in their work, and somehow manage to do so. If they had to tackle something in raw UNIX, they'd be completely lost. I'm afraid that UNIX has a very slow learning curve, and most of these people are already preoccupied thinking about their particular field of work. If you're cloning this and that and mapping genomes, you simply don't have the time to teach yourself to be a UNIX user, let alone a guru. What must you do? 1) Hire somebody, or some bodies, to do it for you. This requires being able to articulate your computer needs in language that your programmer(S) can understand; finding a programmer who has a solid biology background is NOT easy, believe me. I am in fair demand, simply because I'm a biologist who can fake it as a programmer. 2) Ideally, of course, the scientists themselves would do their computer things. In this light, an OS that allows them to do that in the most transparent way is clearly superior (no reference to Tandy intended :-). What OS allows this? Whichever one works for you. No, that's not a cop-out. In our department, we have largely macs, but that is at least partly because our building is wired for AppleTalk networking. The PCs in my lab have ApTalk cards in them to allow access, but it's a kludge (memory distribution is a REAL PROBLEM). In addition, of course, it seems to me that the overwhelming majority of scientific use of computers in my department is for word processing (preparation of papers, grants, etc) and associated graphics, and statistics, and to a growing extent, communications (we now have hooked our network via GatorBox to the campus mainframes...that's how I know how computer- illiterate these people are, I help them log in and the like.). We have found that macs are a pretty robust platform for this sort of thing. There are no unix machines in the department, none, zip. They are prohibitively expensive by and large (no real powerful macs, either; the best are SE/30s). They require computer smarts to set up and operate. They're more trouble than they are currently worth. This may change. As UNIX becomes unified from its current seemingly zillions of incarnations into only a few, as nice clean interfaces are grafted onto it, as prices drop and network access becomes simpler -- why then, people will move that way. Until then, however, there seems little reason to do so, and with the release of the mac's system 7.0 (which I have), several of the nicer features of UNIX are included.... it may be that in five years, all OS's will have pretty much the same features.... That's MY two cents. But, please, let's try to keep this discussion revolving around the use of computers and their OS's in BIOLOGY, okay? Cheers, Josh Hayes, Starving Post-Doc, Zoology, Miami of Ohio jahayes@miavx1.acs.muohio.edu, or jahayes@miamiu.acs.muohio.edu
wrp@biochsn.acc.Virginia.EDU (William R. Pearson) (03/17/91)
Here at U. Virginia, we offered a poorly documented (but functional) set of programs on Unix for several years, as well as the be much better documented more comprehensive Wisconsin package on the VAX. The VAX charges for usage, Unix was free. For this reason alone, unix was far more popular. All incoming graduate students got a 1 hour course on how to use unix for sequence analysis. I don't think that there is a bit of difference between the two systems, or even betwen Unix, VMS, and DOS, for runing sequence analysis programs. (The Mac has the potential to be much better, but program development on the mac is much more demanding) We now have the Baylor MBIR Eugene/SAM packages running on Unix, and people are much happier than before. They also like Eugene/Sam better than GCG, both because Eugene/SAM is free (GCG is not) and because it is menu-driven (GCG requires having the (excellent) manual in hand). Most biologists want functionality over userfriendliness. If your program package is user-friendly, but cannot search libraries quickly, meld sequencing runs, or align multiple sequences, who cares? Today (and for the forseeable future), computers that run UNIX have a price/performance ratio that is about 5X better than either VMS or Macintosh systems. Until more true 32-bit programs become available under Microsoft Windows 3.0 for 386/486 machines, IBM-PC clones, though cheap, are limited. (By the time you add TCP/IP, NFS, and other capabilities that are standard on a Sun or DecStation, Unix is currently extraordinarily expensive on 386/486 machines.) For the sequence analysis program developer, Unix offers the largest set of potential users for sophisticated computationally intensive programs, (and the best programming environment) and thus is the system where the software is developed first. As a result, Unix continues to have the functionality first. People learn to use what they need. Anyone who can design and perform a subcloning experiment - with all the steps required to isolate fragments, match ends, ligate, and transform - can figure out how to login, check a directory, edit a file, and run the sequence analysis programs on unix or VMS. Anyone who is doing a sequencing project MUST learn to the machine. So they do. Bill Pearson