View in EDS

A Method for Detecting Overlapping Protein Complexes Based on an Adaptive Improved FCM Clustering Algorithm

Saved in:
Bibliographic Details
Title: A Method for Detecting Overlapping Protein Complexes Based on an Adaptive Improved FCM Clustering Algorithm
Authors: Caixia Wang, Rongquan Wang, Kaiying Jiang
Source: Mathematics ; Volume 13 ; Issue 2 ; Pages: 196
Publisher Information: Multidisciplinary Digital Publishing Institute
Publication Year: 2025
Collection: MDPI Open Access Publishing
Subject Terms: protein–protein interaction network, protein complexes, fuzzy clustering algorithm, density peaks clustering algorithm, parameter optimization, swarm intelligence optimization algorithm
Description: A protein complex can be regarded as a functional module developed by interacting proteins. The protein complex has attracted significant attention in bioinformatics as a critical substance in life activities. Identifying protein complexes in protein–protein interaction (PPI) networks is vital in life sciences and biological activities. Therefore, significant efforts have been made recently in biological experimental methods and computing methods to detect protein complexes accurately. This study proposed a new method for PPI networks to facilitate the processing and development of the following algorithms. Then, a combination of the improved density peaks clustering algorithm (DPC) and the fuzzy C-means clustering algorithm (FCM) was proposed to overcome the shortcomings of the traditional FCM algorithm. In other words, the rationality of results obtained using the FCM algorithm is closely related to the selection of cluster centers. The objective function of the FCM algorithm was redesigned based on ‘high cohesion’ and ‘low coupling’. An adaptive parameter-adjusting algorithm was designed to optimize the parameters of the proposed detection algorithm. This algorithm is denoted as the DFPO algorithm (DPC-FCM Parameter Optimization). Finally, the performance of the DFPO algorithm was evaluated using multiple metrics and compared with over ten state-of-the-art protein complex detection algorithms. Experimental results indicate that the proposed DFPO algorithm exhibits improved detection accuracy compared with other algorithms.
Document Type: text
File Description: application/pdf
Language: English
Relation: E3: Mathematical Biology; https://dx.doi.org/10.3390/math13020196
DOI: 10.3390/math13020196
Availability: https://doi.org/10.3390/math13020196
Rights: https://creativecommons.org/licenses/by/4.0/
Accession Number: edsbas.BFEEDC6D
Database: BASE
Description
Abstract:A protein complex can be regarded as a functional module developed by interacting proteins. The protein complex has attracted significant attention in bioinformatics as a critical substance in life activities. Identifying protein complexes in protein–protein interaction (PPI) networks is vital in life sciences and biological activities. Therefore, significant efforts have been made recently in biological experimental methods and computing methods to detect protein complexes accurately. This study proposed a new method for PPI networks to facilitate the processing and development of the following algorithms. Then, a combination of the improved density peaks clustering algorithm (DPC) and the fuzzy C-means clustering algorithm (FCM) was proposed to overcome the shortcomings of the traditional FCM algorithm. In other words, the rationality of results obtained using the FCM algorithm is closely related to the selection of cluster centers. The objective function of the FCM algorithm was redesigned based on ‘high cohesion’ and ‘low coupling’. An adaptive parameter-adjusting algorithm was designed to optimize the parameters of the proposed detection algorithm. This algorithm is denoted as the DFPO algorithm (DPC-FCM Parameter Optimization). Finally, the performance of the DFPO algorithm was evaluated using multiple metrics and compared with over ten state-of-the-art protein complex detection algorithms. Experimental results indicate that the proposed DFPO algorithm exhibits improved detection accuracy compared with other algorithms.
DOI:10.3390/math13020196