BMC Bioinformatics

Table 4 Three-level LINDAHL_1.75 fold recognition results of FoldHSphere and FoldHSpherePro in comparison with the state-of-the-art

From: FoldHSphere: deep hyperspherical embeddings for protein fold recognition

Method	Family		Superfamily		Fold
Method	Top 1	Top 5	Top 1	Top 5	Top 1	Top 5
DeepFR (s2) [48]	72.2	85.6	46.6	64.3	50.8	67.1
DeepFRpro (s2) [48]	87.0	93.8	71.9	82.6	63.5	77.2
CNN-BGRU [48]	73.1	88.3	60.1	74.9	60.1	78.7
CNN-BGRU-RF+ [48]	88.5	94.3	74.0	86.3	71.1	86.8
FoldHSphere	77.9	89.0	72.4	87.0	75.8	84.3
FoldHSpherePro	87.9	94.1	81.2	90.0	80.9	88.5

The accuracy (%) results are provided at the family, superfamily and fold levels, considering both the top 1 and top 5 ranked templates. Boldface indicates best performance

Back to article page

ISSN: 1471-2105

Contact us

General enquiries: journalsubmissions@springernature.com