/approximate_ISORANK

Fast and Highly Scalable ISORANK optimization

Primary LanguageJupyter NotebookMIT LicenseMIT

Network Alignment using Approximate-Isorank

This repo contains a faster, scalable implementation of Approximate-Isorank.

Citation:

*** Devkota, K., Cowen, L. J., Blumer, A., & Hu, X. (2023). Fast Approximate IsoRank for Scalable Global Alignment of Biological Networks. bioRxiv, 2023-03. ***

Table of contents

Installation

You can install the netalign package in your local python environment using the command

python setup.py build; python setup.py install;

All the packages dependencies required to run approximate isorank is provided in the environment.yml YAML file.

Running Approximate Isorank

After you install the netalign package, you can run the approximate isorank code by running the command:

netalign isorank --net1 <net1-filename> --net2 <net2-filename> --rblast <rblast-filename> --alpha <alpha-value-float> --niter <no-iterations-int> --npairs <no-of-one-to-one-pairs-int> --output <output-filename>

Parameters

  1. --net1: The PPI network of the first species. Should be tab delimited without a header. Example name: mouse.tsv, fly-string.tsv, mouse-intact.filtered.tsv

  2. --net2: The PPI network of the second species. Should be tab delimited without a header. Follows the same naming conventions as the argument provided by --net1

  3. --rblast: A file that holds the sequence similarity information between the two species. Should be a tab-delimited file and should contain three columns. For example, if the PPI files provided are --net1 fly-string.tsv --net2 mouse-biogrid.tsv, the header of the reciprocal blast file should be fly-mouse.tsv:

          <center>
      
      | fly-string | mouse-biogrid | score | 
      |------------|:-------------:|------:|
      | Q9VPH0     |   Q8R2G6      | xxx   |
      
      </center>
      
      The score should be a floating point number.

  4. --alpha: Isorank parameter, with value between 0 and 1. The recommended value is 0.7

  5. --niter: This parameter decides how close the approximation should be to the true Isorank. If set to 0, this returns the R0 approximation. If set to 1, this
    returns the R1 approximation. If we want to get the true Isorank alignment, set --niter to some value > 10.

  6. --npairs: This parameter decides the number of top one-to-one protein pairs to be outputted by the alignment

  7. --output: The final one-to-one alignment is saved to this output file.

Using Package Functions

We also provide a notebook example showing how Isorank and its approximations can be used directly in code. See the notebooks folder.