I am new to RNA-seq analysis and wanted to ask if there is a method to reliably identify the species of a microbial organism using only its raw short RNA transcriptomics data. Here are some of the possibilities. Please let me know your comments about these methods and if there is some other nice method:

I. Use Kraken2 or MetPhlAn2 to assign taxonomic labels.

II. De-novo assemble the transcripts using Trinity and annotate them through Diamond BLASTx. Then check the most matched genus.

III. Use the genus from the previous case, and find orthologs for its species from OrthoDB. Then do a phylogenetic analysis with the orthologous sequences of all species for that genus, present in my De-novo Assembled Transcripts.

IV. Isolate rRNA sequences from my trimmed reads (~1-5% of the reads), and search in ribosomal databases like SILVA.

Will I get a good alignment if I use the corresponding reference genome for that species from NCBI? Is there a method to merge all reference genomes for that species to account for any non-aligned reads in any one genome?



1 Answer 1


I and IV depending on how the RNAseq is performed. If you are just capturing mRNA then 16S sequence is not retained, admittedly RNAseq has developed.

16S is the traditional taxonomic gold standard in bacteria - phylogenetic a tree of that is definitive. However Kraken2 is an impressive algorithm, just make sure you pick the appropriate database. Kraken2 will classify your bacteria/bacterium against both classified and unclassified genomes. Therein lies its strength. HOWEVER, Kraken2 is not an actual classification, it's simply which genome your bacteria/bacterium will form a sister group with.

The rule of thumb with 16S is <5% genetic divergence is considered a species declaration

This is just a rule of thumb and a Kraken2 analysis will support it nicely.

Thus for bacteria 16S gives a traditional classification and Kraken2 provides a whole genome classification.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.