MinION sequencing details

We followed similar protocols as reported in our earlier bioRxiv preprint (Urban et al. 2015).

There we described our first attempts at sequencing Sciara genomic DNA using modifications to the standard protocol that required no additional reagents or equipment and reported that the majority of our data came from reads > 10 kb with many exceeding 50 kb and a high quality 2D read that exceeded 100 kb. In addition to those early datasets, our MinION data were collected over the course of 18 months and span multiple kits (MAP002-MAP006), flow cell/pore (R7.3 –

(2) Skip the Covaris shearing step to keep DNA long,

(3) Always perform a DNA repair step to repair damaged bases and single-stranded nicks, (4) Use wide-bored tips and very gentle pipetting throughout the protocol,

(5) Use 0.4x ratio of AMPure beads in all clean-up steps,

(6) Add a rinse step before elution of AMPure beads to deplete DNA < 10-12 kb in all clean-up steps, and

(7) Elute DNA off the AMPure beads while adding heat into the system (37-50˚C) for extended periods of time (10-20 minutes) in all clean-up steps.

(8) It is also important to minimize the amount of handling needed, which has been facilitated by updates to the standard protocol by combining End-Repair and dA-tailing into one step.

Reagents, MinION versions, pore models, and protocols for each library:

In total, we prepared 17 libraries for this project. The libraries are named/numbered 01-09, 11-16, 20-21. In general, library preparations have the general workflow of optional shearing, DNA repair, End repair, dA-tailing, adapter ligation, and hairpin enrichment in some cases with various clean up steps in between.

Library 01 used SQK-MAP002 reagents. Libraries 02-08 used SQK-MAP004 reagents.

Libraries 09, 11, 12, and 13 used SQK-MAP005 reagents. All aforementioned libraries used the original MinION and the R7.3 pore model. Libraries 14, 15, 16, 20, and 21 used SQK-MAP006 reagents, the MinION MkI, and the R7.3 70 bps 6mer model. Libraries 14, 20, and 21 were constructed following Oxford Nanopore's standard protocol. Libraries 02-09, 11-13, and 15-16 were constructed using modifications from above as described below.

Optional Shearing:

Libraries 14, 20, and 21 followed the standard protocol, and shearing to ~8 kb was performed in Covaris g-TUBEs as instructed.

For libraries prepared with modified protocols (01-09, 11-13, 15-16), shearing was always skipped and we started out with 2.2-17 μg, using the lower end for early libraries (2.2-5 μg in 01-13) and the higher amounts more recent libraries (15-17 μg in 15-16).

DNA Repair:

All libraries except 01 were subject to DNA repair steps. PreCR (NEB) was performed on libraries 02-09 and 11-13. FFPE Repair (NEB) was performed on libraries 14-16 and 20-21.

End repair and dA-tailing:

For most early libraries prior to MAP006 (excluding 01 and 03), End-Repair and dA-tailing

for library 01. In all others, the tether and motor proteins were pre-combined with other reagents such as the adapters and elution buffer. The first library did not have a hairpin enrichment step, but all other libraries did concurrent with MinION kits and protocols.

Hairpin enrichment:

Seven libraries (02-08) used His-beads (Dynabeads His-tag Isolation and Pulldown; Life Technologies) with MAP004 reagents and four libraries (09, 11-13) used His-beads with MAP005 reagents according to Oxford Nanopore's instructions. The five MAP006 libraries (14-16, 20-21) used MyOne C1 streptavidin beads (Dynabeads/Thermofisher) for hairpin enrichment following Oxford Nanopore's protocol.

AMPure clean up notes:

Libraries 07, 08, 09, and 15 included rinses in all clean-up steps (Urban et al. 2015) whereas the other libraries did not. Clean up steps for all modified protocols used a 0.4x AMPure ratio. For all AMPure beads steps for our modified protocols, DNA was eluted off the beads by incubating at 37˚C for 20 minutes in libraries 01-09 and 11-13, and at 50˚C for 10 minutes in libraries 15-16. For standard protocols, DNA was eluted off beads following Oxford Nanopore’s protocol. Wide-bored tips and gentle pipetting were used throughout.

Sequencing notes:

Sequencing was conducted following standard procedures. For early libraries, smaller amounts of DNA were loaded more frequently (e.g. 4 times) throughout the run. For more recent libraries, half of the library was loaded at the beginning and the second half was added 24 hours in.

Basecalling notes:

Oxford Nanopore’s Metrichor 2d basecaller version 1.10.2 was used for libraries 1-8, version 10.13.1 for library 9, version 1.14.4 for library 11, version 1.19.0 for libraries 12-16, and version 1.20.0 for libraries 20-21.

Performance notes:

Libraries 04, 05, 09, and 11 did not perform well due to flow cells with very limited numbers of pores available for sequencing and consequentially gave little data, but we included them anyway.

Library 11 was attempted on two different flow cells that both started with very few available pores. Moreover, though fragmentation was not performed, the DNA was lower molecular weight than expected as viewed on an agarose gel.

Libraries 06, 07, 08, 09, 12, 13 and 14 were all prepared from the same DNA source to be able to directly compare size distributions after different protocols. Similarly, libraries 15 and 16 were prepared from the same source to make direct comparisons of the effects of including rinse steps in the AMPure clean-ups. When comparing modified protocols without rinse steps to the standard protocols, our modified protocols were enriched for reads >10 kb. Modified protocols that also included rinse steps were additionally enriched over all other protocols for reads >10 kb.

(Q=10.3), 111.2 kb (Q=9.9), 105 kb (Q=9.3), and 100.7 kb (Q=10.4) that aligned across their full lengths with identities of 91.1%, 88.7%, 63.2%, and 84.2%. The 102.9 kb 2D read that we previously reported as high quality (Q=8.74) (Urban et al. 2015) aligned in full at 84.2% identity, representing a high quality, ultra-long 2D read derived from early MinION sequencing reagents (MAP004). For all nanopore reads, 85.7% of 2D reads and 57.9% of 1D reads aligned to a high quality PacBio-only assembly from Canu (Koren et al. 2017). The median percent identities for aligned 1D and 2D reads was 68% and 82.1%, respectively, with both having examples with

>95% identity. Percent identities for both 2D and 1D reads were correlated with mean quality scores (Spearman's rho for 2D = 0.76 and for 1D = 0.70). There were some exceptionally long 1D reads that exceed 200 kb, but those were strictly of low quality and low percent-identity.