Taxonomic assignment discrepancies between SILVA and GenBank databases

pmb0010 · 05-10-2013, 03:20 PM

Hi All-

I have some metagenomic Illumina data that I am attempting to analyze. I am focusing on eukaryotes and have targeted the V9 hypervariable region. I am attempting to assign taxonomy to my representative set of OTU's for a data set. I was using the QIIME-compatible versions of the SILVA 108 release to assign taxonomy (the whole data set and not just the Eukaryote specific data set). I recently attempted to assigning taxonomy of my representative OTU sequences by blasting against NCBI remotely and then importing the blast report into MEGAN. I did both methods for the same data set and same file for the representative OTU's and noticed some discrepancies. Some discrepancies were understandable and were around the genus and species levels. However, other OTU's resulted in very concerning results. One OTU in particular was classified as an Annelid through the GenBank/MEGAN method, but was considered a Chordate when blasted against the SILVA database in QIIME. I would expect discrepancies as you get more specific in the classification (genus/species level) but I did not expect discrepancies so high up in the classification level.

I guess I am wondering if anyone else has attempted to compare or has looked at taxonomic assignments of Eukaryotes using different databases and how well the results matched up. Can anyone suggest why I am seeing such a large discrepancy in taxonomy for some of my samples?

I am fairly new to the metagenomic world and am attempting to figure out how to analyze my data. Any assistance will be greatly appreciated!

rhinoceros · 05-12-2013, 10:53 AM

Have you read the MEGAN manual? Especially the part where it explains how MEGAN assigns reads to taxa? Have you updated NCBI taxonomy in MEGAN? Have you considered that the GI-to-taxid file MEGAN uses is quite old? Hopefully this file will be updated come MEGAN5, or even better, if the author would disclose how the file is processed so we could always have the latest version..

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
pico green discrepancies	schizosaccharomyces	Sample Prep / Library Generation	5	10-17-2013 09:37 AM
Qubit vs Nanodrop discrepancies??	amylee86	General	1	07-10-2013 08:50 AM
tophat / bowtie mapping discrepancies	vineeth_s	Bioinformatics	3	06-10-2011 05:44 AM
Discrepancies between PicoGreen and qPCR in GA libraries	fedora	Sample Prep / Library Generation	3	08-25-2009 06:52 AM

05-10-2013, 03:20 PM	#1
pmb0010 Junior Member Location: USA Join Date: Jul 2012 Posts: 5	Taxonomic assignment discrepancies between SILVA and GenBank databases Hi All- I have some metagenomic Illumina data that I am attempting to analyze. I am focusing on eukaryotes and have targeted the V9 hypervariable region. I am attempting to assign taxonomy to my representative set of OTU's for a data set. I was using the QIIME-compatible versions of the SILVA 108 release to assign taxonomy (the whole data set and not just the Eukaryote specific data set). I recently attempted to assigning taxonomy of my representative OTU sequences by blasting against NCBI remotely and then importing the blast report into MEGAN. I did both methods for the same data set and same file for the representative OTU's and noticed some discrepancies. Some discrepancies were understandable and were around the genus and species levels. However, other OTU's resulted in very concerning results. One OTU in particular was classified as an Annelid through the GenBank/MEGAN method, but was considered a Chordate when blasted against the SILVA database in QIIME. I would expect discrepancies as you get more specific in the classification (genus/species level) but I did not expect discrepancies so high up in the classification level. I guess I am wondering if anyone else has attempted to compare or has looked at taxonomic assignments of Eukaryotes using different databases and how well the results matched up. Can anyone suggest why I am seeing such a large discrepancy in taxonomy for some of my samples? I am fairly new to the metagenomic world and am attempting to figure out how to analyze my data. Any assistance will be greatly appreciated!

05-12-2013, 10:53 AM	#2
rhinoceros Senior Member Location: sub-surface moon base Join Date: Apr 2013 Posts: 372	Have you read the MEGAN manual? Especially the part where it explains how MEGAN assigns reads to taxa? Have you updated NCBI taxonomy in MEGAN? Have you considered that the GI-to-taxid file MEGAN uses is quite old? Hopefully this file will be updated come MEGAN5, or even better, if the author would disclose how the file is processed so we could always have the latest version.. Last edited by rhinoceros; 05-12-2013 at 10:59 AM.