Peak detection, feature extraction and clustering of peptides fragments ions

Koena Monyai, Terence Van Zyl, Stoyan Stoychev

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This work presents a peak detection technique used to detect Proteomics fragments peaks and investigates if shape-based features and clustering can group the peaks such that the clusters are homogeneous, i.e. contain peaks from a single class. We used Continuous Wavelet Transformation (CWT) and two Gaussian Mixture Model (GMM); K=2 and K=15; for peak detection and clustering, respectively. GMM(K=15) performed better than GMM(K=2) with an f1-score of 0.81 and 0.57 for the good class and the bad class, respectively. Additional features and other clustering techniques need to be investigated to improve the homogeneity of the clusters.

Original languageEnglish
Title of host publication2019 6th International Conference on Soft Computing and Machine Intelligence, ISCMI 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages144-149
Number of pages6
ISBN (Electronic)9781728145778
DOIs
Publication statusPublished - Nov 2019
Externally publishedYes
Event6th International Conference on Soft Computing and Machine Intelligence, ISCMI 2019 - Johannesburg, South Africa
Duration: 19 Nov 201920 Nov 2019

Publication series

Name2019 6th International Conference on Soft Computing and Machine Intelligence, ISCMI 2019

Conference

Conference6th International Conference on Soft Computing and Machine Intelligence, ISCMI 2019
Country/TerritorySouth Africa
CityJohannesburg
Period19/11/1920/11/19

Keywords

  • Continuous wavelet transformation
  • Feature extraction
  • Fragments
  • Gaussian mixture model
  • Proeteomics

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Computer Vision and Pattern Recognition
  • Computational Mathematics
  • Modeling and Simulation

Fingerprint

Dive into the research topics of 'Peak detection, feature extraction and clustering of peptides fragments ions'. Together they form a unique fingerprint.

Cite this