Implement bloom filter

[donaldcampbelljr]

Work has successfully begun in this PR: #94
Currently bare bones.

[donaldcampbelljr]

CLI

Added CLI yesterday for basic usage.

CREATE COMMAND EXAMPLE

./target/release/gtars igd bloom --action create 
--universe "/home/drc/Downloads/bloom_testing/real_data/data/universe.merged.pruned.filtered100k.bed" 
--bedfilesuniverse "/home/drc/Downloads/bloom_testing/test1/two_real_bed_files/" 
--bloomdirectory "/home/drc/Downloads/bloom_testing/test1/" 
--bloomname "test" 
--numitems 10000 
--falsepositive 0.001

SEARCH COMMAND EXAMPLE

./target/release/gtars igd bloom --action search 
--universe "/home/drc/Downloads/bloom_testing/real_data/data/universe.merged.pruned.filtered100k.bed" 
--bloomdirectory "/home/drc/Downloads/bloom_testing/test1/" 
--bloomname "test" 
--querybed "/home/drc/Downloads/bloom_testing/test1/query2.bed"

Performance Testing

Did some manual performance testing, would be nice to implement Criterion (https://crates.io/crates/criterion) for testing the permutations of variables including:

• size of inputs files (# chroms, # of regions), size of universe for tokenization, size of query bedfile, number of hash functions, size of bloomfilter, false postive rate

Input files: total of approximately 8.5 million regions
Universe: 100k regions

Creating using Blooms:
~ 16 seconds
~4.4mb
Creating using IGD:
~8 seconds
~140mb

Querying a file with 24 regions:
bloom method:
0.071 seconds
Igd search:
0.009 seconds

Querying file with 7.1 millions regions:
bloom method:
11 seconds
igd search:
16 seconds

Bloom Filter Tree

Began work on implementing bloom filter tree with this commit: c749f21