Sequence That Moldy Lanternfly!

Googling on some keywords from a social media post led to an interesting PNAS article from 2019 on two fungi which can infect the spotted lanternfly.  Lanternflies are a species accidentally imported from Asia which have been striking fear into many plant lovers in the Northeastern US.  If they'd just stick to tree-of-heaven, an imported plant that grows well in the US but accrues little to any love, but instead they suck sap from a wide variety of trees and vines, with particular love for grapevines, many orchard trees. and maples - among others.   They have still been spreading alarmingly up the I-95 corridor - often hitching rides as egg masses on the undersides of cars.  Biological control methods are of great interest - birds and toads apparently will consume lanternflies but equilibrium hasn't been reached.  The PNAS paper, written by researchers from Cornell University, looked at two native fungi which can attack lanternflies.  What surprised me on further searching is that one of these fungal species has no genome assembly at NCBI!!!! 

If they weren't so potentially destructive, spotted lanternflies could be a popular sighting in the wild.  As shown in panel A of the 2019 paper, they are extremely colorful - they look like they are more likely to be toys than serious agricultural pests.  While capable of flight, my personal experience is that they generally hop very short distances.  At a site at a retirement community in New Jersey, I've smashed more than a few after catching them on tree trunks or seeing one hop in the grass.  The numbers can be tremendous - I've seen dozens on tree trunks which have been wrapped with a barrier to block further vertical movement up the trunk.  In Philadelphia, I once saw what must have been hundreds on a building wall near Drexel University.

Figure 1 from Clifton et al, 2019, A pair of native fungal pathogens drives decline of a new invasive herbivore. PNAS 116 (19) 9178-9180  Panels B & C show  Batkoa major infected corpses; D is infection with Beauveria bassiana.

The original paper flagged two fungi, Beauveria bassiana and Batkoa major.  A particularly interesting bit in the paper is that infected cadavers show a very distinctive pattern - Beauveria bassiana infected carcasses are found nearly always on the ground but those with Batkoa major are frequently found up in the trees, webbed into place with hyphae.

I'm pretty sure I explored sequencing Beauveria bassiana at Warp Drive Bio.  Our main focus was Actinomycete bacteria, but we did consider sequencing fungi that might express particularly interesting natural products such as cyclosporine or brefeldin A.  I actually had a little cheese party where I had swabbed the bleu cheese in advance - but plumbing those Penicillium genomes properly would be left to the immortal Cheeseman et al publication.  

At the strain factory I've been part of the sequencing team for a number of novel fungi, which generally assemble very nicely when subjected to nanopore sequencing - despite the challenges of extracting high molecular weight from fungal samples.  In general, only very large or complex repeats defeat fully closed chromosomes.

A poke at NCBI Genomes shows 226 entries for Beauveria bassiana, but not a single one for Batkoa major - indeed, just a single one for the entire Batkoa genus.  And that, for Batkoa obscura,  is a very curious assembly

The assembly statistics give the genome as 56.9Mb, but shattered into 282 contigs with a contig N50 of 1.1Mb.  My first reaction is this must be short reads only - but no, it's PacBio data.  So surely it's just low coverage - nope, 704X coverage!!!  Huh?????

Alas, the raw reads aren't available in SRA.  So I can't play with assembling it myself.  Or generate data on what went wrong.  But a few hypotheses I would pursue if this was my project.

One is what the read length distribution is - were the reads just not very long?  Sequence assembly is unusual in the statistical methods world in that there is a sweet spot for data quantity - it could be that 700+X coverage is introducing too many weird errors that are messing up the assembly.  I've never had a PacBio dataset that deep so don't have experience with this data type, but it can certainly be the case with Illumina data that random or targeted downsampling can improve assembly

Another is what is the nature of any repeats that are breaking the assembly.  I didn't ever have a fungal genome with multiple distinct painful repeats, but there were notable repeats.  Saccharomyces assembles to chromosomes with nanopore reads of only about 8kb - except I've never seen the ribosomal RNA array come out as a single contig.  Some of my yeast-oriented colleagues think it basically can't - that the array is dynamic in a culture and there is no single arrangement.  Our favorite Aspergillus strain has a huge repeat - 120kb or so - which arose during a mutagenesis round in an ancestor. The different copies of the repeat have very few polymorphisms, and even with an ultralong nanopore protocol we never got the main array solved.  Interestingly, one copy is separate from the others and has bridged two chromosomes into a single one.

A more recent paper from the same Cornell group has identified more fungi that prey on lanternflies.  Additional identifications include Cordyceps javanica (39 genomes), Fusarium avenaceum  (16 genomes),  Fusarium fujikuroi (31 genomes), Fusarium graminearum (152 genomes), Colletotrichum fioriniae  (27 genomes).

On the thinner side are Sarocladium strictum (3 genomes), Trichothecium roseum (5 genomes),  and Cordyceps cateniannulata (8 genomes), 

Then there's Fusarium concentricum with only a single genome entry - and it's only short read data so the contig N50 is only 160.3Kb.

There are no genomes for Fusarium falsibabinda.  But worse is Flavocillium bifurcatum  - there are no entries in NCBI genome for the entire genus!  Ditto for Samsoniella sp. - not a single entry.  Clonostachys eriocamporesii and Clonostachys rosea - two different species from another genus with a goose egg for NCBI genomes.

It's hard to keep up with all the genome sequences these days, and the flip side of that is it's hard to know what interesting species haven't been sequenced.  These two papers give a large number of target species that could be very interesting.

These seem like they could be very interesting and approachable student or citizen science projects.  Find some infected lanternfly corpses, do some crude dissection of the hyphae away from the insect body, extract DNA, and send off to Plasmidsaurus, Angstrom, your favorite core facility, or another easy-to-use sequencing CRO. 

That approach might well leave the data "contaminated" with lanternfly sequences and maybe even ones coming from it's bacterial endosymbionts - like many sap-sucking insects the lanternfly relies on endosymbionts to synthesize some essential amino acids the sap is poor in.  If the assembly is essentially a metagenomic assembly, then all the more educational value!  Classify each contig for likely source - that sounds like a fun classroom exercise!

So find some moldering cadavers of lanternflies! They might represent completely novel genome sequencing and interpretation opportunities.