I isolated an environmental bacterium, whole-genome sequenced, obtained a consensus of the genome, and determined the closest phylogenic relative (reference).
How can I determine whether the new isolate is a species or strain with respect to the reference? What should be the percentage difference to call a new species? What programs can I use to generate such percentage?
Thank you
•
written
by
marongiu.luigi
FastANI will quickly calculate the identity between two FASTA files.
https://github.com/ParBLiSS/FastANI
There is no hard consensus what constitutes a new species, but you would have a good case if the identity between your bacterium and its closest relative is < 95%.
You can try classifying the genome using GTDBtk:
https://github.com/Ecogenomics/GTDBTk
It works with GTDB to classify your genome(s) of interest. If the obtained classification is at the species level, it means what you have is similar enough to known bacterial species. If you get a classification only up to the genus level (or any level above it), that will again be a strong indicator that you have a new species.
Thank you, I think the FastANI (now SKANI) solution is the fastest. BTW, I got an ANI of 98%, Align_fraction_ref of 95%, and Align_fraction_query of 99%; shall I consider only the ANI as final result?
Your ANI scores would indicate that you have a strain of a known species. Still, it wouldn't hurt to try GTDBtk classification.
Thank you, but installing GTDBtk is a bit cumbersome (90 to 500 Gb of hard disk, 32 h of downloading on my machine); is there an online engine?
KBase should be able to do GTDB classification. You will have to create an account to use it.
Different species have different inherent mutation rates; check if your species of interest has some published values and compute if your isolate lands itself within the 'normal' variation.
Also check if your species has a MLST to check if your isolate clusters with a known strain (you can use e.g. https://github.com/tseemann/mlst ).
You can also use cgMLST (e.g. chewbbaca) to see how your isolate clusters within a set of given strains (which you can obtain from NCBI).
I ran the MLST analysis against 14 genomes and I got this kind of picture
what would be the interpretation? Thank you
Hey,
I don't think you can interpret this graph with so few genomes.
I'd try to load the created matrix into Phyloviz and check the tree how your samples cluster in regard to the known species.
Log in to answer this question.
