16S classifier: a tool for fast and accurate taxonomic classification of 16S rRNA hypervariable regions in metagenomic datasets

Abstract

The diversity of microbial species in a metagenomic study is commonly assessed using 16S rRNA gene sequencing. With the rapid developments in genome sequencing technologies, the focus has shifted towards the sequencing of hypervariable regions of 16S rRNA gene instead of full length gene sequencing. Therefore, 16S Classifier is developed using a machine learning method, Random Forest, for faster and accurate taxonomic classification of short hypervariable regions of 16S rRNA sequence. It displayed precision values of up to 0.91 on training datasets and the precision values of up to 0.98 on the test dataset. On real metagenomic datasets, it showed up to 99.7% accuracy at the phylum level and up to 99.0% accuracy at the genus level. 16S Classifier is available freely at http://metagenomics.iiserb.ac.in/16Sclassifier and http://metabiosys.iiserb.ac.in/16Sclassifier.

Figure 1. Optimization of parameters using hypervariable region V3.

(a) OOB error at different mtry values for 2-mer, 3-mer, 4-mer, 5-mer and 6-mer models, (b) Effect of k-mer size on time required for the calculation, (c) Size of the input file (used for training) for different k-mer size. From the figure (a), it is apparent that the OOB error for 2-mer and 3-mer models was higher as compared to 4-mer, 5-mer and 6-mer models. The figures (b) and (c) show that the time taken and the training data size were several fold higher for 5-mer and 6-mer models as compared to the 4-mer model.

Figure 2. OOB error shows a slight increase on removing variables.

The optimizations were carried out using hypervariable region V3, 4-mer as input and mtry = 8 (The values of these parameters were selected from the Fig. 1).

Figure 3. Decrease in OOB error for was observed on increasing the number of trees (ntree) at mtry = 8.

This optimization was carried out using hypervariable region V3, 4-mer as input variable, mtry = 8 and 256 variables

Figure 4. OOB error decreases on increasing the number of trees (ntree) at optimum mtry for different HVRs.

For all individual hypervariable region regions mtry value was optimized separately (using 4-mer as input) and was used for constructing the model at ntree = 1000. V2_mtry8 represents hypervariable region V2 at optimum mtry 8, and similarly represented for other hypervariable regions.