Is there a way to use RDP (Ribosomal Database Project) or a similar taxonomy classification tool to produce localized "features" that distinguish 16S sequences based on their phyla? By "feature" I mean something like a region of the 16S sequence that is particularly important in distinguishing between two taxa.

Background: I am comparing different pipelines for the analysis of bacterial 16S sequences. In some cases I see that different read preparation strategies (for example, truncating reads differently) lead to clear changes in the relative abundance of taxa (at the phylum level, for example). I am trying to trace this issue to its cause, and thought that it might help to know which regions of the 16S are especially important in distinguishing two given taxa.