I am doing some bacteria ID work in biological specimens which may contain different bacteria. Most of the time I cannot get pure culture and I just amplify DNA using genus/species specific primers (16s, ITS, etc). However, I am not good at bioinformatics and after getting sequencing results, I usually select the highest score sequence (From BLAST) as the bacteria ID. It work well for specific sequence but if I use genus specific primers, I always get a few possible species. Can anyone suggest ways (based on the DNA sequence) for me to discriminate among them or at least eliminate some options?
What I am doing now is amplifying some conserved sequence and do sequencing. when the results are not specific enough, I tried other DNA sequence, I just feel like I didnt extract all information from the sequence before I move on to use other gene target which waste lot of time. I have seen that some paper can obtain nucleotide pattern (at which position there is always a specific nucleotide and a few of this nucleotide make an specific pattern), I wonder how we can come up with such pattern, if I have to align sequence, like one particular species, how do I select the model sequence for that species(there are a lot different sequencing results from same species) or i need to collect as much data as possible from same species to come up to similar conclusion? Sorry for such long questions, or can anyone suggest some reference reading for me? Thank you!
Alan chan as per my understanding u want to classify the bacterium obtained form samples into particular genus and species.......
So first of all u can target 16S rRNA locus of every bacterium and then after the seqeuncing u can anlyse in into BLASTN program in FASTA format given below
>query
ATTAGGACGGTAGGCGCCA
THen u can choose the database of 16S rRNA availble in program
and it will give u the close homology of the bacterium showing best homology......
For furhter reding u can visit my link for the paper which will give u detail idea of classifying any bacteium into genus and species
Article BLASTN Based Methodology to Classify Benzene Degrading Uncul...
thank you Dilip! I will try to read it.
After blast, how do I select the one with high homology? does that mean the one with highest score? sometime they have 100% identity with 9X% coverage, sometime it got 9X% identity with 100% coverage, Im not sure about the statistic behind, so i shd just stick to the score blast suggest?
see 16S rRNA gene is 1500 base pair and if your query 16 S sequence is completely sequenced i e., 1500 bp then, the subject means which is showing homology might show some variations due to strian variations and it is acceptable for close homology study and hence, the obtained homology (% identity) based results u can consider for the top scored organism and further name it accordingly.......
"Another thing need to consider when u are analysing partially sequenced gene of 16S rRNA"
When u BLAST the sequence, it could be possible that match may come for the another partially sequenced gene previously submitted in database. This happens because u have given the partial sequence and remaining nonsequenced part is not available in ur sequence and hence it will show second best match to completely sequenced gene........
16S rRNA are not always a good target to identify bacteria at the species level. Even with the entire sequence, some genera (or families such as enterobacteriaceae) :
- have very similar sequences among different species.
- have several operons with sometimes quite different sequences.
If using the 16S you find out which genera you want to identify, then switch to the best house keeping genes for those genera. The ultimate approach is MLSA (multi locus sequence analysis).
And in conclusion,
- dont use BLAST, except if the best hit has high % of similarity, covering almost 100% of your query sequence.
- look not only at the best hit, but also the following ones.
AND, retrieve these sequences, align to your query and do a phylogenetic analysis...
AND dont forget, using the NCBI taxonomy you end up with many sequences deposited with a wrong name. Check that these sequences have a pubmed ID, go and read the publication, or use a reference database that has checked annotations !
Thank you Richard. In most case, I can get a high percentage hit (99%) ID, but there are many similar sequence, so I guess I have to do MLSA? For the part concerning phylogenetic analysis, if I have my sequence, I should have retrieve some "model" sequence from genbank for alignment? But for same species, there are always hundred of sequences available, so how do I choose from them? Can you give me more information on doing such analysis? If I produce a phylogenetic tree from the analysis, I just assume the one with the highest similarity is the one I got?
Actually I have read one paper talking about species ID using 16S, but it is rather simple and a bit long time ago. (16S Ribosomal DNA Sequence Analysis of a Large Collection of Environmental and Clinical Unidentifiable Bacterial Isolates)
Hi Alan
DOnt get 16S rRNA sequences directly from genbank, ddbj or embl. Too many are wrongly annotated.
go to Silva or RDP to retrieve these sequences. In fact they also provide tools for identifications. But the best tool is probably http://www.ezbiocloud.net/eztaxon.
you should also connect to http://www.bacterio.net/, the reference for bacterial nomenclature.
I have a quite old tutorial there
http://bioinfo.unice.fr/blast/viroblast/docs/blast_databases.html and also old
http://bioinfo.unice.fr/blast/
Best regards
Richard
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools