Hi!
We are analyzing the bacterial communities of tap water and we ended up with a high percentage of unclassified bacteria. What is the best way to handle them? I Should we treat them like any other taxa? Is it ok to remove them and then normalize all the samples?
Thank you very much!
I've encountered similar results in the past from my studies. What database are you using (i.e. greengenes, silva, rdp or blast)? RDP and BLAST is among the best databases for environmental samples. I suggest a number of ways you can do if you still have numerous unclassified samples.
1. Remove unclassified and focused on the classified (there will still be some relevant information to get from your identified taxa).
2. Check your rarefaction curve. If it is too steep (not making a plateau) just go to genus level or family level classification.
3. Try checking multiple databases for annotation and identification of your sequences.
4. By the way how is your quality filtering? How many reads per sample and how long is the average read length? Just wondering.
I hope these suggestions help. Let me know how you are doing?
Cheers!
What to do with a high percentage of unclassified OTUs? - ResearchGate. Available from: https://www.researchgate.net/post/What_to_do_with_a_high_percentage_of_unclassified_OTUs2 [accessed Apr 7, 2016].
If I am sure that these have been filtered (barcodes removed, etc) and are not chimeric sequences, I would look at the sequences before discarding them. Pull the sequences out, do a database search yourself. Sometimes your cut-off is just too high (probably 97%) for such short sequences that it can't be resolved. You might be looking at a new genus/species/family
PLEASE SEE THE LINK BELOW:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2976220/
- Badges
- Science topic