mapping mutation to a chromosome location with mapreduce/PIG or Disco

Question

Goal: To map mutation location from file1 to a region or feature from file two. For this you need to make sure that chromosome (chr1) and strands (+/-) are the same before comparing chromosome location from file 1 to regions of file2.

Question: How to use mapreduce or Disco to map one location to a region.. . Aka formulate the location -> chromosomal region in a mapreduce method?

Description: I have two medium sized files (10gb) and two file types that I wanted to process. I already have these files parsed in basic python but I will likely have to parse many larger similar files in the future so I wanted to try it with mapreduce (hadoop/Pig to be more specific)or Disco to learn .

While I can run the nodes on an EC2 cluster ideally a one cluster hadoop (yes I know it defeats the purpose) or on something like Disco or Sparc.

I like the idea of using Pig because that would reduce the process to just processing the file from .csv files but I have no idea for how to use mapreduce for mapping something to a region instead of just a key/value pair

Here is a visual representation of what I was thinking of:

File info:

1. First file is TCGA cancer SNP mutations. Some important features include

   • Chromosome location
   • Chromosome number
   • strand
   • sample id
   • the rest is not so important

2. 3' UTR sequence.

   • Chromosome start location: int
   • Chromosome end location: int
   • Chromosome number: chrX
   • strand +/-
   • gene id
   • the rest is not so important

sample files are here:two sample files

Finally python is my language of choice for this if it matters..