TooT-T: discrimination of transport proteins from non-transport proteins.
Amino acid composition
Ensemble learning
Transporter prediction
Journal
BMC bioinformatics
ISSN: 1471-2105
Titre abrégé: BMC Bioinformatics
Pays: England
ID NLM: 100965194
Informations de publication
Date de publication:
23 Apr 2020
23 Apr 2020
Historique:
received:
25
11
2019
accepted:
09
12
2019
entrez:
24
4
2020
pubmed:
24
4
2020
medline:
30
4
2020
Statut:
epublish
Résumé
Membrane transport proteins (transporters) play an essential role in every living cell by transporting hydrophilic molecules across the hydrophobic membranes. While the sequences of many membrane proteins are known, their structure and function is still not well characterized and understood, owing to the immense effort needed to characterize them. Therefore, there is a need for advanced computational techniques takes sequence information alone to distinguish membrane transporter proteins; this can then be used to direct new experiments and give a hint about the function of a protein. This work proposes an ensemble classifier TooT-T that is trained to optimally combine the predictions from homology annotation transfer and machine-learning methods to determine the final prediction. Experimental results obtained by cross-validation and independent testing show that combining the two approaches is more beneficial than employing only one. The proposed model outperforms all of the state-of-the-art methods that rely on the protein sequence alone, with respect to accuracy and MCC. TooT-T achieved an overall accuracy of 90.07% and 92.22% and an MCC 0.80 and 0.82 with the training and independent datasets, respectively.
Sections du résumé
BACKGROUND
BACKGROUND
Membrane transport proteins (transporters) play an essential role in every living cell by transporting hydrophilic molecules across the hydrophobic membranes. While the sequences of many membrane proteins are known, their structure and function is still not well characterized and understood, owing to the immense effort needed to characterize them. Therefore, there is a need for advanced computational techniques takes sequence information alone to distinguish membrane transporter proteins; this can then be used to direct new experiments and give a hint about the function of a protein.
RESULTS
RESULTS
This work proposes an ensemble classifier TooT-T that is trained to optimally combine the predictions from homology annotation transfer and machine-learning methods to determine the final prediction. Experimental results obtained by cross-validation and independent testing show that combining the two approaches is more beneficial than employing only one.
CONCLUSION
CONCLUSIONS
The proposed model outperforms all of the state-of-the-art methods that rely on the protein sequence alone, with respect to accuracy and MCC. TooT-T achieved an overall accuracy of 90.07% and 92.22% and an MCC 0.80 and 0.82 with the training and independent datasets, respectively.
Identifiants
pubmed: 32321420
doi: 10.1186/s12859-019-3311-6
pii: 10.1186/s12859-019-3311-6
pmc: PMC7178945
doi:
Substances chimiques
Membrane Transport Proteins
0
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
25Références
Curr Protoc Bioinformatics. 2013 Jun;Chapter 3:Unit3.1
pubmed: 23749753
Proc Natl Acad Sci U S A. 1981 Jun;78(6):3824-8
pubmed: 6167991
PLoS Comput Biol. 2005 Aug;1(3):e27
pubmed: 16118665
PLoS One. 2014 Jun 26;9(6):e100278
pubmed: 24968309
IEEE/ACM Trans Comput Biol Bioinform. 2016 Sep-Oct;13(5):947-953
pubmed: 26571537
Nucleic Acids Res. 2016 Jan 4;44(D1):D372-9
pubmed: 26546518
Nucleic Acids Res. 1997 Sep 1;25(17):3389-402
pubmed: 9254694
Q Rev Biophys. 2003 Aug;36(3):307-40
pubmed: 15029827
Anal Biochem. 2019 Jul 15;577:73-81
pubmed: 31022378
BMC Bioinformatics. 2013 Nov 28;14:343
pubmed: 24283849
BMC Genomics. 2015;16 Suppl 12:S6
pubmed: 26677931
Nat Biotechnol. 2007 Oct;25(10):1119-26
pubmed: 17921997
Proteins. 2001 May 15;43(3):246-55
pubmed: 11288174