I attended a talk by Josh Quick at PoreCampAU 2017, in which he discussed some common barriers to getting both good sequencing yield and long read length. It mostly boils down to being more careful with the sample preparation. Bear in mind that the MinION will still sequence a dirty sample, it will just be at a reduced yield. Here are my notes from that talk:
- The hardest thing about MinION sequencing is getting the sample in the first place
- There are lots of different sample types and extraction methods
- You can't get longer reads than what you put in; shit in leads to shit out
- DNA is very stable when not moving, but very sensitive to lateral damage
- The phenol chloroform method of DNA extraction is very good, and can be used with a phase-locked gel to make extraction easier
- A simple salt + alcohol extraction might be the best method for extraction (because it involves the least amount of work on the DNA)
- EDTA (e.g. as found in TE buffer, and many extraction kits) is not compatible with the rapid kit
- The most consistently good Nanopore runs produced by Josh's lab were 1D ligations runs on R9.4; the best overall run was a phenol-chloroform extraction + rapid kit
- John Tyson can tune himself out of a low-yield hole (via software)
- Getting small numbers of short reads is very important
- Suggested (and mostly untested) purification techniques: spin column (60-100kb); ethanol extraction (100-150kb), dialysis (150-250kb); low melting-point agarose plug (~1Mb, DNA extraction in situ)
- The nanopore protocol input is in nanograms, but should really be stated as molarity; the kit expects about 0.2 pmol input
- Picture molecules tethered to the surface of the membrane. You can then see that the flow cell density is independent of the sequence length
- Tapestation, Qubit and Nanodrop are all a good idea; a DNA sample that can pass all three tests will work well: no short-length shoulders by Tapestation, sufficient DNA by Qubit, high purity by Nanodrop
- RNA can interfere with sequencing; digesting RNA is recommended
- Freezing DNA is a really bad idea. The ice crystals are very good at chopping DNA up into small pieces
- DNA that is kept in the fridge is remarkably stable; can be kept for over two years (and probably indefinitely)
Update on yield:
We had a run in June 2018 which produced about 3 million reads from PCR-amplified mouse cDNA (read N50 of ~1kb), and that was in a situation where the it was the first sample prepared by someone, the input DNA amount was probably about 20% of what it should have been, and we suffered a dataloss incident. If that's what happened with a bad run, I have high hopes for our runs in the future.
Our next cDNA run in August 2018 produced about 7 million reads and a similar read N50 (i.e. total yield of about 7Gb). The run was interrupted by a Windows update during the first night (I had disabled automatic updates previously, but they had been re-enabled), and there have since been additional software and hardware updates since to improve sequencing yield. If we have a good flow cell and similar sample prep, I expect that the next cDNA run that we'll do should produce over 8 million reads, and possibly over 10 million.