3
$\begingroup$

I am currently an MSc student studying molecular biology (on paper I'm a chemist) and soon I have to talk to my supervisor about my future, PhD research topic. Besides molecular biology I'm also interested in machine learning, mostly in evolutionary algorithms (evolutionary strategy alg.s, genetic alg.s) and Bayesian optimization, from both of which I have written multiple, working programs in C# at home. Due to the massive need for people trained in informatics I'd like to convince my supervisor to include this topic into my future research, especially as I have already acquired some basic skills in these areas. However I have only seen research examples that are purely computational and these didn't include any laboratory work. Can you provide me some insight into the use of these techniques in molecular biology and ways of connecting these with laboratory work?

$\endgroup$

closed as primarily opinion-based by David, kmm, AliceD Mar 5 '18 at 21:21

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    $\begingroup$ Why are you coding in C-sharp? This is a language almost never used in bioinformatics or molecular biology, not least because it is specific for Microsoft Windows, a platform that is generally avoided in this area. Sounds rather as if you are putting the cart before the horse. If your supervisor can't suggest a wet project that requires computational analysis, this list isn't going to. Perhaps you need to find a new supervisor. $\endgroup$ – David Feb 24 '18 at 20:29
  • 1
    $\begingroup$ @David I don't restrict the language options only to C#. I can also program in Python at some level, but I am willing to learn new languages. The main reason I'm asking this is that my supervisor hasn't got much experience with machine learning but he is at home on for example MD simulations, model fitting, etc., and machine learning can come in handy in these areas. I know how to write GAs, I know how to do research in the lab, I know that these two can be somehow used together, but I cannot give concrete examples. It's just I have acquired this skill that I dont want to completely neglect. $\endgroup$ – fazekaszs Feb 24 '18 at 21:05
  • $\begingroup$ OK, but in my experience people tend to do "all informatics" PhDs, or do a PhD where they generate the sort of data (e.g. RNAseq) where it helps if you can use Unix and write the odd glue script, but for the most part you are using powerful programs written by professional computing scientists. I'm not sure it's as easy as some of your respondants make out. $\endgroup$ – David Feb 25 '18 at 18:35
  • $\begingroup$ @David What languages are mostly used for this sort of stuff? $\endgroup$ – Zebrafish Feb 25 '18 at 20:19
  • 1
    $\begingroup$ @Zebrafish — These days everbody in bioinformatics (except oldies like me) seems to use Python wherever they can. The heavy-weight stuff is often written in C++ (e.g. parts of Trinity). Java is still used (because it is taught by CS Departments), but less than a few years ago. Web stuff tends to use MySQL or other relational databases and various JavaScript and other frameworks. PHP is less popular than in other areas. But that's just my impression — others may have a different view. $\endgroup$ – David Feb 25 '18 at 23:17
1
$\begingroup$

There are countless really worthwhile ways to use evolutionary computation (EC) in molecular biology. The chief advantage of EC is that it can find solutions to problems that are too complicated, too nonlinear, etc., to solve by more direct methods. I've used genetic algorithms to evolve models, to find patterns in data, to design optical systems, you name it; and almost always a GA finds solutions I'd never find on my own. Sometimes it's a challenge to find a representation for the solution space that fits the problem well, or to find recombination/mutation operators that work well, but that's where human insight is important. Bottom line: if you can list some problems that fascinate you (or better, fascinate both you and your supervisor), I'd be happy to give you some concrete suggestions.

$\endgroup$
  • $\begingroup$ Our groups research focuses mainly on the structural investigation of miniproteins (such as the Trp-cage motif) through NMR, XRD and CD studies, but we are pretty diverse when it comes to research; my current thesis-work is about bacterial expression optimization in a biorector, we also synthesize small peptides, sugar-derivatives (through classical organic chemical synthesis), do MD calculations and all of these topics are coordinated by my supervisor (so we could say that all of these fascinate him). As for me, I am willing to do almost any research in these areas. $\endgroup$ – fazekaszs Feb 25 '18 at 20:37
  • $\begingroup$ I've opened a chat room, "EC in Molecular Biology", which should be a better place to carry on the conversation. Optimizing bacterial expression in a bioreactor is probably a great place to start looking for an application for evolutionary computation. $\endgroup$ – S. McGrew Feb 26 '18 at 0:23
1
$\begingroup$

The first thing that springs to mind is modeling and simulating the evolution and specificity determinants in protein - protein binding. You can easily include an experimental part and make it as big as you like (the bigger, the better). For a useful set of physiologically important protein - protein interactions and background info on the specificity and affinity determinants in those systems, see Ivanov et al., 2017 and Ivanov et al., 2016, and the references therein.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.