Finally, we validated the effect of database dependency on the accuracy of KO assignment.
In this study, we used the KAAS system for functional annotation to the human metagenomes and also attempted to estimate database dependency on the accuracy of the KO assignment using the E. coli draft genome.
The amino acid sequences of complete CDSs were randomly fragmented to 50, 60, 80, 100, 120, 150, and 200 residues in length, and each fragment was subjected to verification of database dependency based on the accuracy of KO identifier assignment (Additional file 2: Figure S9).
In order to determine the impact of the missing values within the PreVILIG database on the performance and accuracy of the prediction models, missing data were imputed by means of multiple imputation using "Multivariate Imputation by Chained Equations" (MICE) [ 24].
Moreover, applying automated methods to retrieve information from these databases could compromise on the accuracy.
Our findings, as for all database studies, are dependent on the accuracy and completeness of recording in the database [ 38].
The reliability of the records in the CITES database is entirely dependent on the accuracy at which CITES Parties report these data.
However, in steps 2 and 3 on the JVPD database, the accuracy was not increased when training data were added, but it even decreased in step 4. In Table 5, while most of the database has lower accuracy when more classes are added, the accuracy is almost not changed in the TMW and even increases in the ISOLET.
As the database grows, the accuracy of the predictor improves.
A computer-assisted literature search was performed of the MEDLINE, EMBASE and Cochrane databases for relevant publications on the accuracy of MR-colonography in detecting colorectal lesions (see Appendix).
