menu
Channels
# All channels
view-forward
# General
view-forward
# CLI
view-forward
# Online
view-forward
Team

ANI calculations

June 5, 2019 at 7:22am

ANI calculations

June 5, 2019 at 7:22am
Hi, Few months ago I run the MiGA NCBI Prok for two of my genomes. The ANI gave a table with the average and standard deviations quite well. Since I noticed that novel genomes have been added to the database, yesterday I re-run my genomes to see the ANIs between my strains and the recently added ones. However, now the ANI table shows me very different results: although the ANI% between my genomes and previously compared reference genomes are similar, the standard deviation and the fraction of shared genomes are QUITE different... For instance, my genome matched with the following reference before and now: Past ANI (4 months ago): Bartonella tribocorum CIP 105476 NC 010161 89.09% 4.3(SD) 58.66 (shared) Now ANI (1 day ago): Bartonella tribocorum CIP 105476 NC 010161 89.29% 0.0 (SD) 73.86 (shared)
Do you have any idea why these discrepancies? Was the ANI method changed recently?
Thanks!! Ricardo

June 5, 2019 at 8:05pm
Hello Ricardo,
Yes, the method for NCBI Prok was recently changed from BLAST-based ANI (using ani.rb from the Enveomics Collection) to FastANI (see Jain et al 2019). The new method is much faster and the ANI value doesn't change much, but unfortunately it doesn't provide a standard deviation and the shared fraction is only a rough estimate. The old BLAST-based estimate is likely much more accurate for that value (shared). However, the shared fraction for the AAI should still be very accurate. Take a look at the AAI table (under the ANI table, when available), and see the fraction shared for that same match.
Thanks for the question! Please don't hesitate on asking any follow-ups or any other questions.
Best,
  • reply
  • like

June 6, 2019 at 5:47am
Hello Ricardo,
Yes, the method for NCBI Prok was recently changed from BLAST-based ANI (using ani.rb from the Enveomics Collection) to FastANI (see Jain et al 2019). The new method is much faster and the ANI value doesn't change much, but unfortunately it doesn't provide a standard deviation and the shared fraction is only a rough estimate. The old BLAST-based estimate is likely much more accurate for that value (shared). However, the shared fraction for the AAI should still be very accurate. Take a look at the AAI table (under the ANI table, when available), and see the fraction shared for that same match.
Thanks for the question! Please don't hesitate on asking any follow-ups or any other questions.
Best,
Hi Luis, thank you very much for your prompt answer and explanation, now everything is clearer. Indeed, I noticed that the AAI was the same as obtained previously. Thanks again, Ricardo
  • reply
  • like