Tuning back into #NanoporeConf to hear the great @BenedictPaten talking about their human genome work on nanopore - 11 Genomes in 9 days, assembled by SHASTA - very good results (whole MHC in one contig) and with HiC gets you to almost chromosome assemblies. Whoop!
That was then read data a 89% accuracy back last year - now, just software updates (in Guppy) 93% accuracy and pulls in the left hand tail. This gets even better with R10 chemistry - homopolymers well called to 8 or 9 length.
SHASTA has improved a lot, taking NG50 from 19.2MB to recalled nanopore and better shasta to 32.2MB NG50 - with ultra-long and R10 this can go to 58.5 MB (this is bonkers!)
(Just remarkable de novo assembly using nanopore+SHASTA - brilliant).
Real tour de force - many single chromosome arms come out with full single alignments, 99.75% identity and indels are dropping from 520/100KB to 76/100KB (and polymorphisms are in - ie, there is a floor here above 0).
Do tests with DeepVariant Diploid calling GIAB dataset - short indels are worse than illumina but SNVs is *better* - this is due to SNVs in duplicated regions. Where are the issues - homopolymers >=11bp.
(I note that @BenedictPaten hasn't taken the Benito scheme to this data. So again more headroom).
Share this Scrolly Tale with your friends.
A Scrolly Tale is a new way to read Twitter threads with a more visually immersive experience.
Discover more beautiful Scrolly Tales like this.