---------------------------------------------------------------------------
uKIN - Using Knowledge In Networks
----------------------------------------------------------------------------
I. Input
There are three required input files:
1) a network file
2) a prior knowledge file containing a list of nodes (genes) known to be
disease associated, possibly with weights on them
3) a file of newly implicated genes, each with a weight
!!! Plus a required path to a Matlab executable since uKIN needs Matlab to do
some heavy lifting of big matrices. The path should be specified as:
matlab=/path/to/matlab after the three input files.
Additionally, the user may provide:
5) a custom value of the restart parameter alpha controlling how much
influence each type of information (prior and new) has.
Default is alpha = 0.5
6) a custom value of the diffusion parameter gamma controlling how far the prior
knowledge spreads. Default is gamma = 1.
7) output prefix which is used in the beginning of the name of
the output file
II. Output
output_prefix_results.txt is written in the uKIN directory. The file
contains a list of candidate genes ranked by how frequently they are
visited as the guided walks reach the stationary distribution.
III. How to run
Simply issue:
ruby uKIN.rb network_file.txt prior_knowledge.txt new_genes.txt matlab=/path/to/matlab {alpha=0.4 gamma=0.8 output_prefix=my_output}
Note that uKIN is implemented in Ruby but requires Matlab for the heavy matrix
operations.
You can simply install Ruby via: sudo apt-get install ruby-full
Matlab is typically provided for free to academic institutions. You can also
download a 30 day free trial from their website.
IV. Input File Formats
1. Network file: each line specifies an edge, white space delimited:
GENE_ID GENE_ID
2. Prior knowledge: list of genes
GENE_ID
GENE_ID
3. Newly implicated genes: each line is white space delimited
GENE_ID WEIGHT
GENE_ID WEIGHT