Evaluation of the RDP classifier accuracy using 16S rRNA gene variable regions

Abstract EN

The RDP Classifier is a widely used bioinformatic program that performs taxonomic classification of 16S rRNA gene sequences.

However, the accuracy of the program is not clear when it is applied to common PCR products of the 16S rRNA variable regions, which are heavily used in microbiome projects.

In this study, fulllength 16S rRNA gene alignments from the SILVA database were used to simulate the PCR products of the combined variable regions (i.e., V1–V3, V3–V5, and V6–V9).

The classification accuracies obtained from RDP Classifier were evaluated for each of the simulated 16S rRNA regions, at different confidence score thresholds.

Although minor bias was observed, the RDP Classifier achieved overall similar and accurate classification results when using the combined variable regions of the 16S rRNA gene, i.e., V1–V3, V3– V5, and V6–V9.

Additional analysis showed that V2 and V4 were the most accurate among individual regions (i.e., V1 to V9).