Science has always been a social activity. In this tutorial, we will build an application in OCaml to support collaborative research amongst scientists. A classic 3-layer architecture will be used: a domain logic in the middle, a database backend to persist analysis results, and a web interface to present the application's functionality to end-users.
Obtaining data and analyzing it are two basic scientific activities. In the domain layer, we will implement a function to download data from a widely used public genomics data repository, and then use OCaml-R to analyze this data. The downloaded data as well as the analysis results will be saved to a PostgreSQL database using PG'OCaml, which validates OCaml code at compile time against the database schema. Finally, we will implement a website using Ocsigen to summarize the downloaded datasets and visualize the analysis results. Time permitting, we will add user accounts and an access control policy to give end-users the ability to select the collaborators they share their data with.
By the end of this tutorial, participants should feel comfortable building a complete application end-to-end in OCaml. It should be clear how each layer's functionality could be extended to build a full enterprise application. The tutorial should be of interest to non-scientists as well. One needs only to replace tools in the domain layer to build applications for other purposes.
Audience should be comfortable writing basic programs in OCaml. Working through any one of the introductory texts on OCaml is sufficient. Familiarity with the specific libraries being used in the tutorial is not necessary. Their use will be covered in the tutorial.
On September 15th, please download the
virtual machine
needed for this
tutorial onto your laptop. Note the URL will not be active before then.
Install
VirtualBox
if you do not already have it. Open VirtualBox's
preferences and select the Network tab. In the Host-Only Networks box, you
should see one entry, vobxnet0. If not, click the + button to add it. Then,
import the VM into VirtualBox using the File -> Import Appliance command.
During import, you will have the option of changing the VM's settings. Do
not change them unnecessarily, especially the Network settings. It should be
okay to change the name and path to the Virtual Disk Image if needed.
Once the import is complete, boot up the VM and make sure it runs correctly.
The user account name is camel and the password is OCaml with a capital
O and capital C. In particular, make sure you can do the following:
Run ifconfig in the VM, and note the IP address on eth1. From a terminal window on the host, ssh to the VM using this IP address. It should let you connect.
From within the VM, ping to an external site such as google.com. It should work.
is a researcher at the Center for Genomics and Systems Biology and the Courant Institute at NYU. He is designing a domain specific language, implemented in OCaml, for managing and processing large quantities of bioinformatics data in a high performance computing environment. His background is a mix of the natural sciences, with studies in Chemistry and Biology, and computer science, a PhD from CMU in the area of programming languages.