Completed on 22 Jan 2018 by Benjamin Schwessinger . Sourced from https://www.biorxiv.org/content/early/2018/01/11/246280.
Login to endorse this review.
Thanks for posting this preprint. The detail of analysis and the availability of all code is great. it is excellent to see more plant pathogenic obligate biotrophic fungi sequenced. My 'feel' is that these genomes may well look pretty different to some of the better studied non-obligate oomycetes and fungi e.g. 'two speed' genome with effectors clustering to TEs. I could conceived that at least a subset of effectors may well be required in obligate biotrophs as they have to infect the host to complete the life-cycle.
Some thoughts and questions:
- Would be great to see some read length statistics on your PacBio sequencing to get a better understanding why the genome is still in a good number of contigs.
l. 146 Instead of beginning and end of contig I would use 5' and 3' prime of the sequence.
l. 185 ff. I got confused here as the numbers didn't add up for me 6039 single-copy groups give rise to 6,844 one-to-one mappings? I think I get it after reading it several times, yet some rephrasing may well help. Else proteinortho with the synteny flag may have also been an option for doing this analysis.
l. 223ff: The observation of smaller parts of the genome being reshuffled in DH14 vs. RACE1 is pretty interesting. We saw something similar comparing the two haploid genomes in wheat stripe rust fungus (see Figure 2, https://www.biorxiv.org/con.... Wonder how this all happens. Else http://assemblytics.com/ may also be a useful too in future to compare two genomes with each other in regards to structural variations.
l. 265ff: Great analysis on paralogous. We still need to do this for our candidate effectors, yet we saw an overall 'clustering' of candidate effectors. I liked the part of looking if SPs are enriched on certain contigs. Does this also hold true if you consider gene content and not only contig length?
Figure 4A would be easier to interpret if it were normalized to the number of genes analyzed and n given within the figure.
l. 353 ff: Mirrors what we found in wheat stripe rust and others in P. coronata, where candidate effectors do not reside close to TEs in general and not in gene sparse regions. We also see that candidate effectors such as CEPS in Figure S2 C have no really close neighbours. This is pretty intriguing to me. Any thoughts on this? Have you tested if CSEPs are somewhat linked to BUSCOs following the idea that some effectors are necessary in obligate biotrophs. If that is the case for you guys as well, i would be happy to look into if the BUSCOs or effectors tor which this is true are conserved.
l. 380 ff: The analysis of a TE burst in Bgh is very interesting indeed. I think it would profit from a bit more detail on what kinds of TEs were found and how much each family covered. Figure 5 also lacks some details about the usage of all these acronyms used in the figure eg. BOTR? Increasing font size and including a key in the legend would be great.
What I wonder with BGH is where did all the old TEs go? Wouldn't you expect to have some of the older TEs still present around the same age/%id as in the other Blumeria? Within the Blumeria how many TE families were specific to each species? Could it be that your database does not include the most recent TEs from other fungi?
Supplemental figure[:-3]: Not sure that joyplots are the best representation here. A circos plot maybe a better visualization.
Great work. Gave me some good pointers for my own work.