A succinct data structure to store and query colored DBG.
Note : Rainbowfish builds atop the BOSS representation of the dBG, as implemented in VARI. Currently, installing and using Rainbowfish requires first properly installing VARI and then checking out the Rainbowfish repository as a subdirectory of that project.
How to install:
- go through all the steps of installing VARI
- git clone sparsepp into 3rd_party_inst/include
make(in cosmo directory)
- git clone rainbowfish inside the main cosmo directory.
- cd into the rainbowfish sub-directory:
How to run:
- Run all the first four steps in VARI, building KMC2 files and sorting kmers. After that building color matrix using command "cosmo-build"
To compare rb results with VARI you need to run "pack-color" command from VARI too.
- pack colors using the color file created by cosmo and save them in hard disc:
> rb-pack-color kmc2_list_1000.colors 1000 <Address to bitvector dir> <1pass/2pass>
- validating our query of (color & edge) vs VARI's:
> rb-validate kmc2_list_1000.dbg kmc2_list_1000.colors.sd_vector <Address to bitvector dir> <validation-type> > rb-validate
validation-type should be one of the following words :
- compare (to compare results with cosmo),
- query (to go over all pairs of edge/color sequentially),
- cosmo-query (to go over all pairs of edge/color on cosmo data structure),
- random-query (to go over a random pair of edge/color)
- find bubbles:
> rb-find-bubble -a 1 -b 2 ecoli6_63kmc2_list.dbg ecoli6_63kmc2_list.colors.sd_vector \<Address to bitvector dir\> \<type of the bitvector (uuu/uuc/ucc/ccc)\> > rb-bubbles
type of compression shows for each of the three bitvectors (label, boundary, and equivalence table) if the bitvector is compressed (c) or uncompressed (u), so for example ucc means label bitvector is uncompressed, boundary is compressed, and equivalence table bitvector is also compressed. As in paper, we use ucc.