This program aims to find out the essential proteins in the Protein Network and evaluate its accuracy by comparing with prepared reference data.

As a sample, we use 5 algorithms which are shown below:

• Betweenness Centrality (BC)
• Closeness Centrality (CC)
• Degree Centrality (DC)
• Eigenvector Centrality (EC)
• Significance-Based Essential Protein Discovery (SigEP) *

And 4 data sets:

Network	Proteins	Interactions	Density
DIP	5093	24743	0.0019
MIPS	4546	12319	0.0012
Krogan	2674	7075	0.0020
Gavin	1430	6531	0.0064

As a result of various experiment, it indicates that there is no "best" algorithm but a "better suitable" one.

Details are shown in the last part.

Help

Use '-h' to show:

===== Center Protein Calc =====

Usage: [option] <value>
-h See this.
-i Specific input data path.
-r Specific reference data path.
-s (optional) Specific step (default 1).
-a Use 5 centrality algorithms together (BC, CC, DC, EC, SigEP).
-b Use algorithm Betweenness Centrality (BC).
-c Use algorithm Closeness Centrality (CC).
-d Use algorithm Degree Centrality (DC).
-e Use algorithm Eigenvector Centrality (EC).
-p Use algorithm Significance-Based Essential Protein Discovery (SigEP).

=========== Caution ===========

Must have at least one of ['-a', '-b', '-c', '-d', '-e', '-p'].
Must have '-r' and 'your refer data's path'.
Must have '-i' and  'your input data's path'.

============ Tips ============

Algorithm BC or CC will take a long trip to run (like O(N^3)), but it works!
Use '-b -c' together (save you 50% time)

============ About ============

Author: bipy@GitHub
Version: 20200725.1

BC and CC use "floyd" algorithm to calculate all the vertex, this precedure might take some time.

Luckily they share the same result of floyd, so use '-b' and '-c' together and gain a tea time : )

*About SigEP

Introduction

SigEP can identity the essential proteins from Protein-Protein Interaction network: we present a p-value calculation method for quantifying the statistical significance of each protein by considering both its degree and local clustering coefficient. To reduce the computational cost, we further present an upper bound of the p-value, which is less timeconsuming in practice. After calculating the p-value for each protein, we control the FDR of identified essential proteins using the well-known BH algorithm.

Reference

Liu, Y., Liang, H., Zou, Q., & He, Z. (2020). Significance-Based Essential Protein Discovery. IEEE/ACM Transactions on Computational Biology and Bioinformatics.

Calculate

Where Ci repersents the local clustering coefficient of the i th node, which is defined as:

d is the degree of the node and e is the number of edges among the d neighbors of the node.

And beta, the upper bound:

Build

Requirement: G++ 6.5.0 or higher (No third party library included)

Command:

g++ Essential-Proteins.cpp -std=c++14 -o Essential-Proteins.exe

Usage

Run DC and EC!

Essential-Proteins.exe -d -e -r "Reference essential proteins.txt" -i "original dip.txt"

Run 5 algorithms together!

Essential-Proteins.exe -a -r "Reference essential proteins.txt" -i "original dip.txt"

Run 5 algorithms together and specific step!

Essential-Proteins.exe -a -s 5 -r "Reference essential proteins.txt" -i "original dip.txt"

Result

DIP

MIPS

Krogan

Gavin