big wide table

I have a dataset with ~100,000 columns and ~200,000 rows (GWAS study
data(1))
none of the fields have more than about 15 characters

Sample1,A

I want to be able to extract subsets of columns and rows for analysis

I can see two ways I can approach this:-

Convert it to something RDF like ie

sample_id
column_number
data

to make a table with 3 columns and 20 billion rows

or use the array datatype

sample_id
array_of_data

(1) http://en.wikipedia.org/wiki/Genome-wide_association_study

--
Michael