Hacker News Reader

Comment by bipson | Hacker News Reader

I am (kind of) in research on the CS side, and I wish I was on the medical side.

The grant hunting, bullshitting, old-stuff-in-new-covers is all the same, but the problems and the topics are just so much more approachable, relevant, interesting, and the community seems to be "nicer".

But then of course, I know people that are in that area an were getting into research, and they did not paint a nice picture. And the community obviously can suck also once you're on the inside. And being a "researcher" without clinic duties is really not easy getting into in Europe it seems. Also, who wants to stop doing clinic completely, when this is what he trained for so long...

There are upsides and downsides to everything - I think it really depends on where your interests are.

View on HN

I've done both. The crucial difference is that computer work is far more productive than bench work. If you spend a month at your computer, you will get something out of it. You can easily spend a month at the bench and come away with nothing at all.

Part of the secret to success in bench work is finding a seam which you can mine productively. You have some technique or system set up and working, and you can grind out findings from it. But it takes time to establish that seam, and it isn't necessarily anywhere near where the actually interesting problems are.

I worked on cell migration. We used a particular cell type that had very good, consistent migratory behaviour, which made it a good system. But it was a primary culture, made fresh from dissected tissue every day. Primary cultures are highly resistant to having DNA put into them. Pretty much all modern cell biology hinges on putting DNA into cells. I spent a year or so trying different techniques (including making adenoviral vectors!). Eventually, i found something which worked okay. But that was just the start of actually doing interesting work - the year up to that point had produced nothing. This is completely normal in bench work, but would be pretty shocking with computers.

I've done both too, but after a couple months at the bench, I always find myself spending most of my time trying to create robots to automate the work instead. Haven't gotten that to work out in the long run, so I just do programming now.

I guess at one point computing was more like that, in that to solve a new problem you may well have had to build yourself a new kind of computer or compiler first.

> the community seems to be "nicer"

Having worked as a programmer for molecular neurobiologists for two years: yeah, they are. Everyone's just hyped at how insanely fast things are developing. Everything I saw when I worked there felt cutting-edge. The engineering involved, the science produced, the speed at which new insights were being gained.

> And being a "researcher" without clinic duties is really not easy getting into in Europe it seems.

The trick is to go the molecular biology route instead, it seems

I'm a structural biologist and I can attest that we absolutely need more computer scientists in this field. The problems we are trying to solve are complex and there's so much room for improvement.

By the time someone is an expert on biology and then physics, there's not much mental capacity left for CS. And unfortunately people that start CS and move to biology are extremely rare.

What are the areas of research that need CS people and how would a grumpy CS/engineering MS/PhD make the switch? Um, asking for a friend!

As an undergraduate I actually first got into computing through a biology-computational project! My current work pays the bills, but my armchair dream switch to research - in biology or another science.

Chiming in as a former neuro-y person:

Really, it's the bare basics that they need a big hand with. In bio, a lot of the grad students have literally no calculus or math training outside of stats. They have no training whatsoever in CS, let alone programming. If they are programming, it's matlab, and then they use 15+ nested for-loops (true story). Hardcoded is the only way of things. Github is a way around gmail's 25mb limit. The PIs are these people with a few years under their belts and even worse habits. It's all hacked together with bubblegum and if-statements. And the time/money pressure is intense. They don't have the resources to start fresh with new habits or to learn three semesters of math.

I don't really know where to start on helping them, but it's pretty far back in the the chain. Honestly, given the constraints, it feels like they need someone to write programs for them in a pair coding style for free.

Reply on news.ycombinator.com