Oxford Nanopore Technology (ONT) sequencing enables telomere-to-telomere complete genome assembly because of its potentially unlimited maximum read length. However, ideal read length observed from mammal or yeast DNA are not reproducible with DNA from some other eukaryotes. Drosophila are important model organisms but the community suffer from making long reads with ONT. People usualy get 15 to 25 kb of read N50 with current R10.4.1 Q20+ platform. The best N50 of D. melanogaster in publications is 28.3 kb (Bernard Kim, et. al., PLOS Biology, 2024). Another source sayid their N50 values were 30 kb and 39.9 kb. No real telomere-to-telomere genome assembly is available for now. This research did not assemble any centromeres. Their N50 (LSK109 kit) was 48.2 kb after size-selection by BluePippin system.
I optimized the entire workflow to improve the read N50 to 40+ kb or 70+ kb with the ligation sequencing kit (LSK114) or ultra-long sequencing kit (ULK114) respectively. Neither BluePippin nor SageHLS was used. This is the best know results so far. Part of the results were presented in the 67th Annual Drosophila Research Conference, 2026. I am aiming at the read N50 greater than 100 kb. All experiments were performed on a PromethION P2 Solo machine.
The sample was D. pseudoobscura. The throughput was 23GB in 20 hours on an used flowcell.
The sample was the most common species, D. melanogaster. The throughput was 45.2GB in 20 hours and then 28.2 GB in 16 hr on a brand new flowcell.
To my best search, there is no public available records of ONT ultra-long data of Drosophila so far.
This is the result from the latest experimental protocol. The sample was D. melanogaster.
This is a typical result with the current stable workflow. The sample was D. pseudoobscura.
This is a typical result with the current stable workflow. The sample was D. virilis, whose genome contains 50% of heterochromatin and repettive satellites.
It was performed side-by-side with another Drosophila melanogaster sample with the same protocol. The sequencing result at that time is:
In the past year, I found the bias of ONT sequencing which "hates" some kind of repetitive motifs. Depending on the kit (ULK114, LSK114, NBD114...), the coverage also fluctuate across the genome. By "leveraging" the bias and loooooooog reads, I just had a automatically complete genome of a common Drosophila species (Let's keep this species a secret and we are rushing on analysis.) Known telomere-specific TEs were detected in the ends of contigs (and we suspected there are few more lowkey).
