A systematic evaluation of nucleotide properties for CRISPR sgRNA design

BMC Bioinformatics

Table 3 AUC, Youden index (J), Sensitivity (Se) and Specificity (Sp) from inter-platform comparison (training set: ribosomal and non-ribosomal genes, test set: mESC essential genes)

Feature class	AUC	J	Se	Sp
PD Mono	0.797	0.486	0.751	0.735
PD Dinuc	0.832	0.544	0.792	0.752
Freq	0.751	0.382	0.716	0.667
Align	0.574	0.131	0.490	0.641
Thermo	0.641	0.261	0.817	0.444
Packer	0.667	0.241	0.514	0.726
PhyChem	0.726	0.351	0.718	0.632
PseKNC	0.733	0.370	0.660	0.709
Comb Feature	0.848	0.566	0.843	0.722
azimuth	0.795	0.463	0.857	0.607
sgRNA Scorer	0.669	0.288	0.548	0.739
azimuth (retrained)	0.833	0.543	0.787	0.756
sgRNA Scorer (retrained)	0.804	0.474	0.786	0.688

Comb Feature: PD Mono+PD Dinuc+Freq+Align+Thermo+Packer+PhyChem+
PseKNC. azimuth and sgRNA Scorer were the results based on the softwares by [7] and [27], respectively developed using different training datasets. azimuth (retrained) and sgRNA Scorer (retrained) were the results obtained by refitting the algorithms on the current training set (ribosomal and non-ribosomal genes)

ISSN: 1471-2105