breakage-points

Breakage Point Homework Assignment

Objectives

1. Parse the Tab separated file. Each line specifies a "translocation".

2. Figure out the two most likely FragileSites that yielded a given Translocation and the coordinates of these Fragile sites. It is not required to do this for the entire dataset.

3. Serialize the resulting set of Fragile Sites to something easy to parse. YAML (via PyYaml) and CSV (document your field names) via the built in lib are both good, basic choices.

General Advice

There are lots of ways to do this, and I want to see your approach, but here is a hint from the DB authors: ''' More advanced users might prefer to run customized BLAST searches and inspect the output in order to identify the precise breakpoints, which is the process we followed in the original BMC Genomics paper. This has the advantage that it will uncover microhomologies present at the fusion boundary (there are quite a few of those). ''' I want to see your choice and why. For example "Used BLAST API because I already have a good client for it" or "Wrote out FASTQ file and then invoked bwa-mem as a subprocess because it is fast"

Bonus Points - de-duplicate the set of resulting FragileSites based on their coordinates. That is: if two FragileSites have the same genomic coordinates, then they must be the same fragile site

Equally, you should demonstrate how you know you've found reasonable genomic coordinates. This means you should write an automated test. For example you could assert that the coordinates occur in the stated gene for a sample of results.

It is not important to do this for the entire data set. I don't actually care about the actual output or even speed. I care about your design judgement, if you write clean, PEP08 compliant Python, how you tested it to show it both runs and produces reasonably correct output, and how you document and comment it so that other team members can understand.

If you're not up to date on professional Python coding style check out some popular libraries, such as SQLAlchemy, or "Code Like a Pythonista"

See http://python.net/~goodger/projects/pycon/2007/idiomatic/handout.html

Code quality (readability, consistency) is not emphasised much in academia, but a lack of it drives professional developers insane (in a way that is probably unhealthy). Most developers aren't biologists and will have no idea if your implementation is biologically accurate (that is your concern!), but they will be able to assess its logic and consistency. Issues with code style and readability will present a roadblock to that.

ALWAYS make sure you use the if __name__ == '__main__' idiom in your scripts. I can't tell you how many candidates forget this, causing their code to execute on import.

There are many tools such as PyLint that will check this for you automatically. Most IDEs have this built in. PyCharm has a free community edition and a free or steeply discounted Educational edition that works well.

Data Model

We shall define a Translocation object as a type of GenomeFeature composed of 2 FragileSite objects (which are a subclass of GenomeFeature) such that fragile_site_a + fragile_site_b --> translocation_a_b

A GenomeFeature is an object that has at least

1. a name and
2. various other attributes and
3. coordinates (in a genome)

Coordinates are a tuple of (chromosome, absolute_start, absolute_end, strand). For example (1, 123, 456, -1) would be a Feature that occurs on Chromosome 1 from 123 to 456 on the negative strand, such that it's sequence is reverse_compliment(chromosome_one.sequence[123:456]), that is you can use Python string slicing to get the positive strand sequence, but need to reverse compliment that sequence if the feature is on the negative strand to get its sequence 5' to 3'

Scaffold

A basic scaffold for a Python project has been provided in this repository. It is recommended that a virtualenv is used for your development environment. The project can be installed in "development" mode by running the following command (whilst the virtualenv is activated) in the root of the repository:

$ python setup.py develop

This will install the module brkpoints and any dependencies listed in the setup.py file, after this the test suite can be run with the following command:

$ nosetests

To see more verbose output and the output from print statements, invoke with the following flags:

$ nosetests -v --nocapture