Workflow and Indels in amplicon contigs

[chytrids]

Hello, I am using Canu v1.8 on the Flux Linux-based operating environment at the University of Michigan to assemble contigs of demultiplexed amplicons generated from fungal genomic DNA using a Nanopore MinION. Depending on the species of fungus, the amplicon can range from 4.5 kbp to 6kbp. After filtering and demultiplexing, I have been following this workflow:

1. Attempting default settings:

   canu \
   -p $myprefix -d $myoutputdirectory \
   genomeSize=5000 -nanopore-raw $myfastafile \
   useGrid=false "canuIteration=1" canuIteration=1

2. If (1) failed, adding:

   corOutCoverage=200 correctedErrorRate=0.15

3. If (2) failed, adding:

   corMhapSensitivity=normal

4. If (3) failed, modifying (3) to:

   corMhapSensitivity=high corMinCoverage=0

Would this be a recommended strategy? When moving to steps (3) or (4), should I be removing the parameters added in (2)?

The only problem that I can see from contigs.fasta files produced at different steps in the workflow is that there seems to be a handful of indels that differ between them. Indel differences are also noticeable when comparing our contigs to reference sequences. Is there a recommendation for obtaining the best quality sequences in consideration of indels?

Thank you for your assistance and attention.

[skoren]

I think this is a reasonable approach, you don't need to remove parameters from step2 in 3 & 4. They should be compatible.

As for consensus, you need to run a polishing tool to improve the indel rate, either medaka or nanopolish should do a reasonable job.