Bluesky now has over 10 million users, and I was #2,334,006!

So the real data (just 10% random alignments, i.e., not all the data) to a hifiasm human haploid assembly looks like this with 500-bp pseudo-hi-c contacts

So as I thought, the data are clean, on a test dataset, but I did not think it would be this clean...

So I was surprised that this worked, but it appears that there is rather low read level phase-switching as a result of error-correction of #Nanopore r10.4.1 reads (N50 17kbp) using Herro, followed by Brutal Rewrite, followed by 2x Peregrine_2021 overlapping, then Dechat (thanks #JuliaLang) teaser...

New script (same BAM file as bsky.app/profile/jpel...) but now with 2D contours of average read quality score and read length changes in #nanopore sequencing over time in #JuliaLang
gist.github.com/jelber2/f22c... I know the right (Z axis) scale changes for the contours...

Was able to make an animated GIF of "hourly" changed in distribution of average read quality from #nanopore reads using #JuliaLang gist.github.com/jelber2/547e...

Playing around with racon for overlap-based error-correction, brutal-rewrite for kmer-based error-correction, and peregrine-2021 for overlap-based error-correction on Human Ultra-long Nanopore Simplex reads called with dna_r10.4.1_e8.2_400bps_sup_v4.3.0 on dorado v0.5.0. Duplex yield was super low.