Ultra-fast sequence clustering from similarity networks with SiLiX

BMC Bioinformatics

Table 2 Comparison of clustering performances of SiLiX and hcluster_sg (used alone or in combination with SiLiX).

method (%identity)	Jac.	Spec.	Sens.
SiLiX (0.25)	0.69	0.73	0.95
SiLiX (0.35)	0.78	0.88	0.89
SiLiX (0.45)	0.74	0.94	0.78
SiLiX (0.25) + hcluster_sg⁽¹⁰⁰⁾	0.77	0.85	0.92
hcluster_sg	0.76	0.84	0.91

Specificity, sensitivity and Jaccard coefficient were estimated by comparison with InterPro classification. The best quality metrics are indicated in bold. SiLiX was used with an alignment coverage threshold set to 80% and with various sequence identity thresholds. hcluster_sg was used on the entire dataset or only on the largest families (more than 100 sequences) retrieved by SiLiX with a permissive sequence identity thresholds of 25% (hcluster_sg⁽¹⁰⁰⁾).

ISSN: 1471-2105