How to analyze wrongly sequenced data?

[joshuakwon]

Hi everyone,

I know the tile "wrongly sequenced data" is weird.
There was miscommunications between sequencing service provider and me.
I ordered the PlexWell Pro library prep and sequencing service from seqWell.com (http://seqwell.com/products/plexwell-service/) to perform 16s rRNA sequencing.
I thought they are offering a sequencing service for variable regions of 16S rRNA gene (V3 or V4).
Therefore, I asked questions several times that they can provide 16s rRNA sequencing and which of variable regions they use in the 16S rRNA sequencing.
After the quote, I sent bacterial genomic DNAs from my samples to the company.
Some weeks later, they sent back the sequencing data and told me like this.
----------------------
"I believe there has been some confusion as to what our service and library preparation is and what we deliver. We do not make amplicons from the 16s region.

The plexWell library preparation is an enzyme-mediated fragment library preparation. Thus your sequencing results include all parts of the starting amplicon. I have attached a paper on how to use the data generated from this for 16s sequencing."

Journal of Microbiological Methods 122 (2016) 38–42, "Studying long 16S rDNA sequences with ultrafast-metagenomic sequence classification using exact alignments (Kraken)"
----------------------
I didn't send them amplicon but they seems to put my samples like amplicon into their enzyme-mediated fragment library preparation.
Therefore, I'm not sure what type of sequencing data I've got and I don't know how to handle the data.
In the paper they provided me, they performed amplification of 16S rRNA gene using universal primers, but they didn't do the amplification in my understanding.

With my trouble, I put the sequencing data into the MG-RAST server and the data could be successfully assembled.
The MG-RAST announced me that the data includes the mixture of WGS and amplicon.
After the pipelines of the MG-RAST, I've got some statistics like this.
------------------------------------
Upload: bp Count 5,714,293 bp
Upload: Sequences Count 17,004
Upload: Mean Sequence Length 336 ± 91 bp
Upload: Mean GC percent 44 ± 9 %
Artificial Duplicate Reads: Sequence Count 450
Post QC: bp Count 5,528,556 bp
Post QC: Sequences Count 16,482
Post QC: Mean Sequence Length 335 ± 91 bp
Post QC: Mean GC percent 44 ± 9 %
Processed: Predicted Protein Features 14,222
Processed: Predicted rRNA Features 1,200
Alignment: Identified Protein Features 4,797
Alignment: Identified rRNA Features 63
Annotation: Identified Functional Categories 3,137
------------------------------------
Actually, I prefer more rRNA features than protein features.
However, there are only 63 of rRNA features and I'm not sure the number is suitable for OTU based analysis.
Moreover, I don't know what type of sequencing data I have, I can't trust the results from the MG-RAST also.

In my case, what do you suppose to me to handle the data?
Please, help me to deal with my problem.
The sequencing data is not my own, but it belongs to my customer.
Very many thanks to all of you.

[GenoMax]

Therefore, I'm not sure what type of sequencing data I've got and I don't know how to handle the data.

Sounds like you received whole genome sequence data for your samples (are the individual samples pure cultures or mixed communities). There is not much you are going to be able to do if you were purely interested in looking at 16S. Assembling the data likely gave you mixed assemblies if these samples were not pure cultures.

Mistakes happen and sounds like this one is going to be an expensive one (unless your customer understands the issue and error). Find a different sequence provider and repeat the experiment.