Thread by @Amandatron89

Lilo: A pipeline to stitch together tiled amplicons from Nanopore sequencing data.

github.com/amandawarr/Lilo

But Amanda, that pipeline already exists.

Yes, I know that, me, but the genomes I spend my days arguing with have specific features that make them incompatible with the @NetworkArtic pipeline.

The artic pipeline is based on aligning reads to a reference and polishing the reference to look like the reads. That works great, but...

Pick a Porcine Reproductive and Respiratory Syndrome Virus (PRRSV), any PRRSV (of either species) and throw it against BLAST. This is what you will see. PRRSV has a hypervariable region that in any given sample can be drastically different from all but its closest relatives.

Minimap2 can handle aligning a big amplicon across the gap, but polishing tools curl up into a little ball and cry (I've been there, polishing tools). They are expecting the reference to be constructed from the reads and aren't prepared for this nonsense.

The artic pipeline works great on PRRSV, if you are using a reference that happens to be incredibly similar to what you are sequencing. Otherwise it doesn't finish running.

For African Swine Fever Virus (ASFV), the virus that has gone panzootic and is knocking on the door of the USA, small, medium and large indels contribute more to its diversity than SNPs. sciencedirect.com/science/articl…

Polishing a reference to match reads is likely to miss some of these larger indels. And we really need reliable ASFV genomes right now. journals.plos.org/plospathogens/…

Lilo only uses a reference to assign reads to an amplicon and to order and orient polished amplicon sequences. It polishes a read representative of the amplicon instead of polishing a reference and stitches the polished amplicons together.

Primer schemes for ASFV and PRRSV-1 & -2 will follow soonish. If you have a bunch of ASFV samples you are desperate to sequence, drop me a message.

Lilo does a nice job on PRRSV and ASFV, and has also been tested on SARS-CoV-2 and didn't do anything daft (let me know if it does). Give it a reference, primer scheme and primer sequences and it should adapt pretty well to anything with reasonably sized overlaps.

But will it produce identical SARS-CoV-2 genomes to the artic pipeline?

Well, me, probably, but the depth filters work in different ways, within sample SNPs can be real and homopolymers still cause havoc. But yeah, with good coverage they should be more or less the same.

Me, myself and I are the only ones to have used it so far, so here's hoping it behaves for anyone else who wants to use it 🤞

I wish you the best of luck getting Hawaiian Roller Coaster Ride out of your head #SorryNotSorry

Now with its own logo courtesy of my awesome sister @Jammyllama @UnicornMarCo, who also did the @ION_BRU logo.