I'm identifying plant sequences (for rbcL gene) obtained from insect gut contents using BLAST in the NCBI GenBank database. Most of the sequences have 95-96% of matched identity and a few have of 80-85%. I realize that some plant sequences (as with animals and microorganisms) are not deposited in GenBank so 99-100% match is not always possible. Still I'd like to match the obtained sequences to species/genus or at least family. Also, the sequences I use are of good quality so mismatch is a rare possibility.
I'm wondering if there are any implicit guidelines I can follow to justify my sequence identification for species, genus, and family. Any published reviews or research articles? Couldn't find any so far.
Thank you very much. I would appreciate any suggestions.
Hello @Alina. May be this could help you go through.
There is not a universally applicable quantitative method to make assessment for species, but a good guide involves using four components of the reference sequences in BLAST results:
1: high Identity value, in the ‘identity’ column;
2: high "Query Cover" value, in the ‘Query’ column;
3: the number of reference sequences being matched this high in the results; and
4: that the species name comes from a trusted source.
Furthermore, as a general rule species (query sequence) with >97% similarity could be assigned to same species. However there may be other species in which 99% simiarity are not applicable for species determination, as they may be from two different, related species.
Now coming to your query, let me put it this way:
For the purposes of the initial triage, a high Identity value generally falls in the range of 98-100% sequence similarity. The identity reports on the percentage of base pairs that are the same between the sequence of your specimen and that of the reference specimen. If 99 out of 100 base pairs match, then you have a 99% identity value in your results. If there are a large number of reference sequences that fall into the 98-100% range in your results, all with the same species name you believed it your specimen to be, then you would likely identify your specimen as that species, and not need to review that sequence much more in the future. If more than one reference species have 98-100% sequence similarity with your specimen, you would identify your specimen conclusively at the genus level, but not at a species level.
https://mycoflora.org/index.php/resources/faq/41-examining-your-blast-results
Thank you.
Hi Alina,
I basically agree with Azhar, on principle, but the significance of the % identity depends on the locus. I don't work with plants, so I have no experience with this particular gene, but it seems to be a rather slowly evolving gene, and hence maybe not the best choice for identifying plant species from gut content. If you look up a sequence of the rbcL gene from a random species and then BLAST it, you will probably find (at least I did; try Paeonia veitchii as an example) that there are several other species listed that differ from the query sequence by only 0,5% or so. Therefore, I would say that you need a 100% match to be sure you have the exact species, and anything above around 99% similarity may be in the same genus, but the limits are fuzzy and probably vary between genera.
I suggest that you BLAST your sequence, look at the list given by BLAST and investigate the best match. Retrieve the sequence of the best match, BLAST it, and see what the distance is to the next species, genus, etc. How does your sequence compare to that?
Best
Urban
Urban, thank you very much! This is very helpful approach, I've never thought of BLASTing the sequence from the best match. Yes, my understanding is that rbcL is a slowly evolving gene often used in plant phylogenetics. Thanks again, I appreciate your time.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics