# Workflow of metabarcoding analysis

For a diet analysis of an insect-eating animal, all species in a sample shall be identified. For this a sequencing of the metagenomic sample was done, where the COI/COX region was used as a barcode and amplified. This resulted in 20 million reads (after quality control).

While looking for an apropriate workflow, I am faced with these issues:

• For many tools 20 million reads are to much to handle (BLAST for instance)

• For the majority of metagenomic tools the focus is on micro-organisms, not insects

• The database to search against must contain COI/COX data, mainly from insects, however, the animal could also eat snails or frog eggs...

• Since the reads are amplicons some analysis are not recommended

Leaving me with these questions:

1. What are the typical analysis steps in a workflow like this?
2. Which tools could I use for my purpose?
3. Where could I get a database of COI/COX data?

I'm working on something similar as a side project. I'll detail the steps that I'm using, but I'm curious if others can suggest a better way.

• You don't mention if your data are paired-end, but in my case I have paired-end sequences that overlap in the middle. I used FLASh to merge the reads and seqtk to convert the reads to fasta format.

• I used the cluster_fast algorithm in vsearch to cluster sequences at 100% identity, then filtered out rare sequence types (<10 sequences) to account for sequencing errors.

• This resulted in a manageable number of unique sequences, which I searched against a custom database using vsearch, but which could also have been searched against the nr database using blast.

Regarding your third question ("Where could I get a database of COI/COX data?"), you can apparently search and download fasta sequences from BOLD.

However, I don't see a "select all" button to select all records found by one search in a single fasta file.

There seems to exist an R API: https://cran.r-project.org/web/packages/bold/bold.pdf which refers to this: http://www.boldsystems.org/index.php/resources/api#sequenceParameters>.

From this I seem to understand that one can directly query using urls:

wget -O Gastropoda.fa http://www.boldsystems.org/index.php/API_Public/sequence?taxon=Gastropoda&marker=COI


And I get a fasta file containing COI sequences for Gastropoda. There may be ways to refine the query geographically, using the geo parameter.

• Yes, I already downloaded some of the BOLD data. I circumvented the missing "select all" button by using a firefox addon for selection boxes. I did not know that there is an R API, that will be helpful. Jul 4 '17 at 9:26