Daily Life in an Ivory Basement

Laurie Dillon just posted the SIGPLAN eduction board article on Why Undergraduates Should Learn the Principles of Programming Languages to our faculty mailing list at the MSU Computer Science department. One question that came up in the ensuing conversation was: what functional programming language(s) would/should we teach?

I mentioned OCaml, Haskell, and Erlang as reasonably pure but still pragmatic FP languages. Anything else that's both "truly" functional and used somewhat broadly in the real world?

thanks!

--titus

Our sequencing analysis course ended last Friday, with an overwhelmingly positive response from the students. The few negative comments that I got were largely about organizational issues, and could be reshaped as suggestions for next time rather than as condemnations of this year's course.

The 23 students -- most with no prior command-line experience -- spent two weeks experiencing at first hand the challenges of dealing with dozens of gigabytes of sequencing data. Each of the students went through genome-scale mapping, genome assembly, mRNAseq analysis on an "emerging model organism" (a.k.a "one with a crappy genome", lamprey), resequencing analysis on E. coli, and ChIP-seq analysis on Myxococcus xanthus. By the beginning of the second week, many students were working with their own data -- a real victory. Python programming competency may take a bit longer, but many of them seem motivated.

If you had told me three weeks ago that we could pull this off, I would have told you that you were crazy. This does beg the question of what I was thinking when I proposed the course -- but don't dwell on that, please...

The locale was great, as you can see:

One of the most important lessons of the course for me is that cloud computing works well to backstop this kind of course. I was very worried about the suitabiliy and reliability and ease of use, but AWS did a great job, providing an easy-to-use Web interface and a good range of machine images. I have little doubt that this course would have been nearly impossible (and either completely ineffective or much more expensive) without it.

In the end, we spent more on beer than on computational power. That says something important to me :)

The course notes are available under a CC license although they need to be reworked to use publicly available data sets before they become truly useful. At that point I expect them to become awesomely useful, though.

From the scientific perspective, the students derived a number of significant benefits from the course. One that I had not really expected was that some students had no idea what went in to computational "sausage", and were kind of shocked to see what kinds of assumptions us comp bio people made on their behalf. This was especially true in the case of students from companies, who have pipelines that are run on their data. One student lamented that "we used to look at the raw traces... now all we get are spreadsheet summaries!" Another student came to me in a panic because they didn't realize that there was no one true answer -- that that was in fact part of the "fun" of all biology, not just experimental biology. These reactions alone made teaching the course worthwhile.

Of course, the main point is that many of the students seem to be capable of at least starting their own analyses now. I was surprised at the practical power of our cut-and-paste approach -- for example, if you look at the Short-read assembly with ABySS tutorial, it turns out to be relatively straightforward to adapt this to doing assemblies of your own genomic or transcriptomic data. I based our approach on Greg Wilson's post on the failure of inquiry-based teaching and so far I like it.

I am particularly amused that we have now documented, in replicable detail, the Kroos Lab MrpC ChIP analysis. We also have the best documentation for Jeff Barrick's breseq software, I think; this is what is used to analyze the Long Term Evolution Experiment lines -- and I can't wait for the anti-evolutionists to pounce on that... "Titus Brown -- making evolution experiments accessible to creationists." Yay?

There were a number of problems and mistakes that we had to steamroller through. In particular, more background and more advanced tutorials would have be great, but we just didn't have time to write them. Some 454, Helicos, and SOLiD data sets (and next year, PacBio?) would be a good addition. We had a general lack of multiplexing data, which is becoming a Big Thing now that sequencing is so ridiculously deep. I would also like to introduce additional real data analyses next year, reprising things like the Cufflinks analysis and whole-vertebrate-genome ChIP-seq/mRNAseq a la the Wold Lab. I'm weighing adding metagenomics data analysis in for a day, although it's a pretty separate field of inquiry (and frankly much harder in terms of "unknown unknowns"). We also desperately need some plant genomics expertise, because frankly I know nothing about plant genomes; my last-minute plant genomics TA fell through due to lack of planning on my part. (Conveniently, plant genomics is something MSU is particularly good at, so I'm sure I can find someone next year.)

Oops, did I say next year? Well, yes. If I can find funding for my princely salary, then I will almost certainly run the course again next year. I can cover TAs and my own room/board and speakers with workshop fees, but if I'm going to keep room+board+fees under $1000/student -- a practical necessity for most -- there's no way I can pay myself, too. And while this year I relied on my lovely, patient, and frankly long-suffering wife to hold down the home fort while I was away for two weeks, I simply can't put her through that again, so I will need to pay for a nanny next year. So doing it for free is not an option.

In other words, if you are a sequencing company, or an NIH/NSF/USDA program director, interested in keeping this going, please get in touch. I plan to apply for this Initiative to Maximize Research Education in Genomics in September, but I am not confident of getting that on the first try, and in any case I will need letters of support from interested folks. So drop me a note at ctb@msu.edu.

Course development this year was sponsored by the MSU Gene Expression in Disease and Development, to whom I am truly grateful. The course would simply not have been possible without their support.

My overall conclusion is that it is possible to teach bench biologists with no prior computational experience to achieve at least minimal competency in real-world data analysis of next-generation sequencing data. I can't conclusively demonstrate this without doing a better job of course evaluation, and of course only time will tell if it sticks for any of the students, but right now I'm feeling pretty good about the course overall. Not to mention massively relieved.

--titus

p.s. Update from one student -- "It's not even 12 o'clock Monday morning and I'm already getting people asking me how to run assemblies and analyze data." Heh.

So, I've been teaching a course on next-generation sequence analysis for the last week, and one of the issues I had to deal with before I proposed the course was how to deal with the volume of data and the required computation.

You see, next-generation sequence analysis involves analyzing not just entire genomes (which are, after all, only 3gb or so in size) but data sets that are 100x or 1000x as big! We want to not just map these data sets (which is CPU-intensive), but also perform memory-intensive steps like assembly. If you have a class with 20+ students in it, you need to worry about a lot of things:

• computational power: how do you provide 24 "good" workstations
• memory
• disk space
• bandwidth
• "take home" ability

One strategy would be to simply provide some Linux or Mac workstations, with cut-down data sets. But then you wouldn't be teaching reality -- you'd be teaching a cut-down version of reality. This would make the course particularly irrelevant given that one of the extra-fun things about next-gen sequence analysis is how hard it is to deal with the volume of data. You also have to worry that the course would be made even more irrelevant because the students would leave the course and be unable to use the information without finding infrastructure and installing a bunch of software and then administering the machine.

While I enjoy setting up computers and installing software and managing users, I'm clearly masochistic. It's also entirely besides the point for bioinformaticians and biologists - they just want to analyze data!

The solution I came up with was to use Amazon Web Services and rent some EC2 machines. There's a large variety of hardware configurations available (see instance types) and they're not that expensive per hour (see pricing).

This has worked out really, really well.

It's hard to enumerate the benefits, because there have been so many of them ;). A few of the obvious ones --

We've been able to write tutorials (temporary home here: http://ged.msu.edu/angus/) that make use of specific images and should be as future-proof as they can be. We've given students cut and paste command lines that Just Work, and that they can tweak and modify as they want. If it borks, they always just throw it away and start from a clean install.

It's dirt cheap. We spent less than $50 the first week, for ~30 people using an average of 8 hours of CPU time. The second week will increase to an average of 8 hours of CPU time a day, and for larger instances -- so probably about $300 total, or maybe even $500 -- but that's ridiculously cheap, frankly, when you consider that there are no hardware issues or OS re-install problems to deal with!

Students can choose whatever machine specs they need in order to do their analysis. More memory? Easy. Faster CPU needed? No problem.

All of the data analysis takes place off-site. As long as we can provide the data sets somewhere else (I've been using S3, of course) the students don't need to transfer multi-gigabyte files around.

The students can go home, rent EC2 machines, and do their own analyses -- without their labs buying any required infrastructure.

Home institution computer admins can use the EC2 tutorials as documentation to figure out what needs to be installed (and potentially, maintained) in order for their researchers to do next-gen sequence analysis.

The documentation should even serve as a general set of tutorials, once I go through and remove the dependence on private data sets! There won't be any need for students to do difficult or tricky configurations on their home machines in order to make use of the tutorial info.

So, truly awesome. I'm going to be using it for all my courses from now on, I think.

There have been only two minor hitches.

First, I'm using Consolidated Billing to pay for all of the students' computer use during the class, and Amazon has some rules in place to prevent abuse of this. They're limiting me to 20 consolidated billing accounts per AWS account, which means that I've needed to get a second AWS account in order to add all 30 students, TAs, and visiting instructors. I wouldn't even mention it as a serious issue but for the fact that they don't document it anywhere, so I ran into this on the first day of class and then had to wait for them to get back to me to explain what was going on and how to work around it. Grr.

Second, we had some trouble starting up enough Large instances simultaneously on the day we were doing assembly. Not sure what that was about.

Anyway, so I give a strong +1 on Amazon EC2 for large-ish style data analysis. Good stuff.

cheers, --titus