Skip to main content
. 2018 Jul 31;20(6):2009–2027. doi: 10.1093/bib/bby065

Figure 5.

Figure 5.

Performances of different tools on mouse data set. LncFinder achieved the best accuracy of 0.9347, while PLEK (re-trained model) had an accuracy 0.8178. The accuracy of CPAT (re-trained model) was 0.9242 and better than CNCI’s 0.9133, CPC’s 0.8678 (web server) and CPC2’s 0.8611. LncFinder (Without.SS) outperformed other tools with an accuracy of 0.9286 even without secondary structure-derived features. When using the model for human, LncFinder outperforms CPC/CPC2, CPAT (web server), CNCI and PLEK (re-trained model) with a satisfactory accuracy of 0.9186. Under this circumstance, LncFinder can even rival CPAT (re-trained model for mouse), which demonstrates LncFinder’s robustness and high cross-species stability.