I work in a lab (academic/genetics). We have a public facing website so when we develop tools they go there. So we're using: perl, java, R, python, php apart from the stuff we compile from c.

Stringy code thats hard to decipher can be written in any language I have discovered.

My take, Perl is kinda on the way out, though it was used extensively. BioPerl is a nice package.

Biologist like R, it pretty quick and behaves like they aren't programming. It can make graphs nicely.

Python seems to be the go to compromise. BioPerl is a really nice package. But people start to want to use it for big things and to get it performing adequately requires a lot. There is confusion from the researchers that scares them away: between python 2/3 numby, pypy. BioPython packages are pretty excellent.

If it needs to be fast and crunch large sets of data (fairly common) the tool is in C or C++. We should start using Rust..

We use Php (Silex) to deliver a front end and some quick database lookup and display. Its replacing perl for this.

I've been in the field since 2001. I have seen exactly two perl files in that time

It really depends on the subfield. Sequence-based genomics is very highly Perl-dependent even today. BioPerl is pretty much the standard library for that even though BioPython, etc. are beginning to take over. Other subfields such as differential sequence abundance which are more about the statistics rather than the raw sequence tend to be R-centered thanks to Bioconductor.

I think by your definitions there are sub-subfields then ;) In my neck of the woods sequence based stuff has java & c doing a lot of the heavy lifting.

>modestly more readable languages.

This is definitely all relative, I would argue lisp is easier to read given that the philosophy of its syntax is that there is no syntax...

Though if you've only ever read classical languages I can see how lisp would seem alien

Lisp someone else wrote is harder to read than Perl you wrote, but any Lisp is easier to read than Perl someone else wrote.

My experience with Perl was that me-a-week-ago counts as someone else. That's not flippant — it's an accurate reflection of my experience that Perl tends to be write-only.

There are some people who do write beautiful Perl, but they are few and far between (again, in my experience).

> many users are still using Perl. It would be [hard] to have them migrate

I spent a year as a staff member of a genomics institute, and researchers occasionally came to me for help getting the local sui generis tooling stack set up and working sanely. (Well, I say "stack"; jwz's bookcase-from-mashed-potatoes metaphor hastens to mind.)

Having seen the miseries they went through, I don't think it would necessarily have to be all that hard to make migration look good, especially to a language whose syntax is other than wildly irregular and whose performance is other than usually abysmal.

Java and .NET are also used a lot.

Python more (though I have had to refactor completely incomprehensible Python written by bioinformacians).

Depends on the company.

We have some customers with zero lines of Python code, but they do use R and Tableau.

All the software used to talk to the devices and do ETL processing or graphical analysis is then written in Java or .NET, depending on the department and their set of OSes.