David W Mohr, Ahmed Naguib, Neil Weisenfeld, Vijay Kumar, Preyas Shah, Deanna M Church, David Jaffe, Alan F Scott
Table 1 -- re-ordering the columns in a logical progression from left-to-right would scan better -- so DISCOVAR, 10X Supernova 1.1, BNG+Supernova 1.0, BNG+Supernova 1.1 (and a bit odd that Supernova 1.0 is omitted)
Table 2 takes a lot of space and isn't really giving more information than Figure 5 -- it would be preferable to have plots like Figure 5 for more chromosomes or a table listing the number of scaffolds & the sizes of scaffolds for each chromosome
The fact that the BNG data greatly reduced the number of scaffolds but had only a modest effect on N50 should be discussed. Is this a limit on scaffolding through centromeres? Do any scaffolds appear to cover an entire chromosome arm? Do any cross a centromere? It might be useful to discuss the known chromosome structure of pinnipeds as described in Beklemisheva 2016 -- adapting their Figure 4 to show how your scaffolds relate to human-seal and dog-seal synteny blocks would be valuable.
Figure 4 -- what region is this? Citation for the fact it is a breakpoint in many genome comparisons? Does this map to a known join vs. human karyotype as described in Beklemisheva? This point would be interesting to see discussed.
p.7 "acrocentric human chromosome which are" -- should identify which acrocentric chromosome(s) are being referred to.
It would be of interest to the genomics community to have a histogram of estimated fragment lengths based on the 10x read clouds and the observed lengths of BNG fragments. It would also be useful to have statistics on anomalously-mapping reads -- those that map outside the scaffold to which the majority of the cloud's reads are assigned. A histogram of number of reads per UMI might also be interesting.
Figure 1 -- move the legend into the plot by labeling the lines-- much easier to read, particularly for the colorblind (red vs. green is never a good choice for that reason)
Discovered a small hitch in one thing I suggested -- the Baikal seal in the Beklemisheva analysis has 2n=32 but Hawaiian monk seals have 2n=34 (Lu et al 2000). According to Arnason 1974 the 2n=34 karyotype is probably ancestral with a single fusion generating the 2n=32 karyotype. Fronicke et al 1997 would make the fused chromosome "S", which is homologous to human chromosomes 17 and 5. Some more musings over on the blog
Satomi Mitsuhashi, Kirill Kryukov, So Nakagawa, Junko S Takeuchi, Yoshiki Shiraishi, Koichiro Asano, Tadashi Imanishi
Figure 4b: please make ines much heavier & cinder avoiding color for distinguishing stages - different shapes would work better (e.g. rectangle, hexagon, oval) for individuals with color perception issues.
If you are going to claim portability, then a complete list of equipment is needed, with weights. What are power requirements? Refrigeration requirements?
What is the yield from each run? What fraction of reads in each were unclassifiable? What is expected sensitivity in a more complex sample?
Discussion and legend for 4b should emphasize that you started with purified DNA, not bacteria. So time to lyse&purify DBA would need to be added to running time, and differential lysis/extraction could shift your sensitivity.