Interview with Sandra Reuter

Today I’m talking to Sandra Reuter from the University of Freiburg in Germany. Sandra was a bench biologist, but she’s moved towards bioinformatics. Hi, Sandra. Hi, Martin. Can you tell us a little bit about your background? So yeah, when I started off my PhD, I thought I was going to go to the lab and do lots of mutants in the bug Yersinia enterocolitica. And some of the strains had been sequenced. But it turned out back then that the sequencing wasn’t complete, so I was asked to look at sequences and sit in front of a computer for six months. And I thought, that’s quite challenging. Six months is a long time. I want to go back to the lab.

That was nine years ago– no, even longer, now 12 years ago– and I’ve never really left the computer, because it was so fascinating. And now I work at the University of Freiburg at the Institute of Infection Prevention and Hospital Epidemiology. And what I do in my day job, I guess, is sequencing potential transmission chains to look whether people have transmitted bugs between them or whether there’s an outbreak on the neonatal intensive care unit– for example. I’m also interested in antibiotic resistance, so also looking into highly-drug resistant bugs at the hospital. So Sandra, how would you advise people to start if they’re moving from the bench to the computer?

I think a good place to start is always looking for courses to attend, like Wellcome Trust Advanced Courses, or some other online resources. I think that’s a good place to start to get some overview and just the basics to get on with. But then, afterwards, I think the main thing I would suggest is having a problem that you can apply the command line to, or with any course, really, that you learn, if you go to an R course or whatever, it’s always having, then, a problem that you want to apply it to and actually put it into practise.

So a course is only the start, and then I usually find that once you put it into practise, you actually find out whether you understood what you’re doing. So yeah. So it’s really having the application of what you’re doing? Yeah, pretty much. So there’s no point in just attending classes because you’re interested, because by the time you want to actually apply it, it might be years down the line, and then you’ve forgotten half the things– and the field has moved on as well. So yeah, I think having an application in mind already is helpful when trying to look at courses. So having data, say, that you need to work on, and then going from that. Yeah, mm-hmm. OK, that’s great.

When it comes to working on this, how would you solve problems? Oh, I Google a lot. [LAUGHS] I usually tend to take my error messages and just put them into Google and see what happens, because there’s quite a lot of help out there. Even if I’m stuck programming or whatever, I try and put my problem into Google, and there’s quite useful web pages out there. I think one’s called Stack Overflow. Some are specific for sequencing and bioinformatics. They usually come up as quite high hits in Google. And then, I just click through and try different things and try. So when I read answers, I guess I look for, do I understand what they’re doing?

Because I’m not just going to type some random thing in there if I don’t know exactly what the client’s doing. But if I think, yeah, I understand where this is going, I think this looks OK, I’ll try it, I might need to try several different approaches. But yeah, usually, that helps a lot, just Googling. I guess if you have a colleague next to you that knows these kind of things, it’s more helpful. But if you’re on your own, Google is your best friend. I think I tend to find the same. It’s very much Google the problem and then see where that goes.

You normally get an array of answers from people, and it’s difficult to choose, but try and choose the one that’s the most simple, I guess, is always the– That as well yeah. So what have you found most difficult about moving from the bench to the ground lab? I guess it’s a different way of communicating with a computer and looking at things. Just like the structure of Unix, you can’t click anywhere. You just have to type a lot of stuff. That is different. And it’s sometimes little things that drive me crazy– like you always have to watch where you put your spaces, and have you finished a semicolon at the end of the line, or whatever?

It’s usually these things that cause problems. And I guess it’s a bit like learning another language, really, because the computer has a very structured way of how you need to communicate with it. But once you know the basics, it’s pretty easy to build out from there. Exactly, yes. Yeah, I find the typos and things that you put in there– Oh, god. If somebody looks over your shoulder, they find them instantly. Yeah, yeah. Tab complete’s my best friend, I have to say. And so the middle mouse button, just this copy and pasting things, just to get rid of all the spelling mistakes that you eventually make. Those are key things, yeah.

I think initially, when you’re learning, then you don’t just want to cut and paste things into the window, because you don’t learn what the command is doing. Mm-hmm. Once you know, it’s safer or easier to Tab Complete and to cut and paste where possible. Yeah, especially filenames or whatever. Oh, yes, spaces– spaces are dangerous. No spaces in filenames, ever. Yes, yeah, and you find that converting files from, say, Windows or Mac to Unix, where spaces in the filenames are fine, it’s a little tricky when you do that. I guess the question is, would you do the same thing again? Would you move from the lab to more of a computational– Yeah. Yeah, very much.

It’s fun, once you get down to it. It’s a slightly different way of working, I guess. Because what I found frustrating or challenging at the beginning, I guess, was there was this other PhD student working next to me, and at the end of the day– so she was working in the lab; I was with mainly on the computer. And at the end of the day, she had a bucket full of consumables and empty Petri dishes and stuff to show the amount of work that she’d done. Whereas, I could just say I’ve added some bytes to my files, but it wasn’t tangible.

So I found it challenging that with bioinformatics, sometimes, you have to work for quite a long time before you actually see some results, before you can make a nice figure. Because I was doing annotations, so it took about half a year before I could finally do some genome comparisons, and find out what was shared and stuff, and then produce something. So at the beginning, that took quite a long time, just to get used to that. But having said that, it’s good fun, because in the lab, also, things don’t always work. Bacteria don’t grow, or there’s contamination, or an experiment might also take quite a long time and then not work, and you have to start again.

So I quite like what I’m doing now, and the way things have developed, bioinformatics and microbial genomics is really important. So it’s the future. It’s important to know about it, to learn about it to some extent, at least. I mean, things might get more automated down the line, but I still think it’s important to learn about it and be able to apply and judge the data. So I would definitely recommend it, yeah. Yeah, I think that’s– what you mentioned, you touched on automation there. And I think it’s great, initially, to be able to do the things individually. But then, you want to either automate things yourself or working in an Institute where you have a team that can automate things.

That’s also great. And also, if you– I mean, there’s lots of commercial software out there. But I still think with a commercial software, because sometimes it’s a black box, if you don’t know what you’re doing, you also don’t know where things go wrong. So it’s always useful to know some of the basics so that you’re able to judge whether things look all right. Yeah, for example, I once used this web page to look for MLST genes, and all of a sudden, it was giving really weird results– like only six genes were reported, but it was putting out an ST.

And I only found out because I knew that those ST numbers were far too high, because I’d picked some low numbers. But in the clinic, a random ST number might be quite plausible. So yeah, it’s good to take these kind of things, to be aware of what you’re doing, what kind of results you should be getting, and also knowing what good data looks like. I do put a lot of emphasis on QC for my sequencing data, so that I’m only working with stuff that’s good, because your analysis, your results and decisions in the clinic might depend on good data. So– Thank you, Sandra. That’s been fascinating. You’re welcome. It’s been nice to chat to you. Thank you for watching.

Please leave any questions and comments on this video in the Comments section.