The PFP and ESG protein function prediction methods in 2014: effect of database updates and ensemble approaches

GigaScience

Table 4 Examples of predictions by CONS and individual-component methods. Capsid protein (UniProt ID: Q9IVZ8)

Method	GO id	Confidence score	GO term
CONS	GO:0019028	1.00	viral capsid
	GO:0005198	0.97	structural molecule activity
	GO:0019012	0.70	virion
	GO:0039615	0.68	T = 1 icosahedral viral capsid
	(GO:0032774)	0.43
	GO:0003723	0.43	RNA binding
	GO:0044228	0.43	host cell surface
	GO:0030430	0.43	host cell cytoplasm
PFP	GO:0044228	1.00	host cell surface
	(GO:0032774)	1.00
	GO:0030430	1.00	host cell cytoplasm
	GO:0005198	1.00	structural molecule activity
	GO:0003723	1.00	RNA binding
	(GO:0006351)	0.71
	GO:0043656	0.65	intracellular region of host
	GO:0033646	0.65	host intracellular part
	(GO:0008150)	0.59
	GO:0003676	0.59	nucleic acid binding
ESG	GO:0019012	1.00	virion
	GO:0019028	1.00	viral capsid
	GO:0039615	0.99	T = 1 icosahedral viral capsid
	(GO:0019048)	0.15
	(GO:0030683)	0.15
	(GO:0039573)	0.15

GO terms in bold are correct annotations of the protein. Terms in italic indicate parental terms of correct GO terms. Terms in parentheses are wrong predictions
For CONS prediction, GO terms that have a confidence score larger than 0.4 are listed. For PFP prediction, GO terms that have a confidence score larger than 0.5 are listed. For ESG, all predicted GO terms are shown

ISSN: 2047-217X