BMC Research Notes

Table 1 Most important nucleotide attributes that were selected by different weighting algorithms

From: Prediction of hepatitis C virus interferon/ribavirin therapy outcome based on viral nucleotide attributes using machine learning algorithms

A
Subtype 1a (Responders vs.		Subtype 1b (Responders vs.
Non-Responders)		Non-Responders)
Attribute	No. of selective	Attribute	No. of selective
	attribute		attribute
	weightings		weightings
	(out of 10)		(out of 10)
Count of hydrogen	9	Count of GC	8
Count of oxygen	8	Count of UA	7
Count of CA	7	DS Count of nitrogen	7
Count of CG	7	Count of AU	6
Count of Cytosine	7	Count of GG	5
Count of Guanine	7	Count of Uracil	5
Count of GU	6
Count of UU	5
Count of UA	5
Count of CC	5
B
Subtype 1a (Responders vs.		Subtype 1b (Responders vs.
Relapsers)		Relapsers)
Attribute	No. of selective	Attribute	No. of selective
	attribute		attribute
	weightings		weightings
	(out of 10)		(out of 10)
Count of oxygen	10	Count of UU	6
Count of UU	7	Count of CA	5
Count of Uracil	7	Count of carbon	5
Count of nitrogen	6

Ten algorithms (PCA, SVM, Relief, Uncertainty, Gini Index, Chi Squared, Deviation, Rule, Information Gain, and Information Gain Ratio) were used to determine the most important nucleotide attributes for the prediction of HCV subtypes 1a and 1b responders from non-responders (A) and responders from relapsers (B). Common nucleotide attributes used for genotypes 1a and 1b have been bolded. A: adenine, T: thymine, C: cytosine, G: guanine.

Back to article page

ISSN: 1756-0500

Contact us

Submission enquiries: bmcresearchnotes@biomedcentral.com
General enquiries: ORSupport@springernature.com