Adventures with the MinION Part III: Optimizing Protocols and Technology For Success
As mentioned in previous posts (found here and here), optimization of this immature technology is one of the major hurdles users face. In order to help others bypass the learning curve, we have laid out some of the problems we’ve faced and the solutions we’ve found. Our lab’s primary goal for using the MinION is whole genome sequencing and de novo assembly of yeast mutants.
Library Prep Optimization:
Producing enough DNA output from the ID2 prep kit was one of our initial problems . The protocol suggested an input of 1-1.5 ug of DNA, which was prepared through multiple AMPure bead wash steps, adapter ligations, and the addition of various other reagents (if you are part of the Nanopore Community, find the protocol here). The expected output is 200ng. In spite of adhering closely to the protocol, we were generating around 50-100 ng according to PicoGreen analysis. Unfortunately, this is not enough DNA to warrant using valuable flow cell sequencing time (flow cells function for 48 cumulative hours of sequencing before they ‘die’, more on this later). By scouring the Nanopore Community, we found many other users were having similar problems.
These were their suggestions:
-double the input DNA
-don’t elute from AMPure beads as often
-keep the DNA in the same tube as beads during ligation
-incubate AMPure beads at 37C for 30 minutes during elution
-45 min incubation to bind DNA to beads
After following these recommendations, we started to acquire adequate output. From an input of 2.5 ug and longer incubation and elution time, we saw outputs of 450-700ng.
Optimizing Library Prep for Long reads:
In addition to the previous preparation problem, we ran into an issue with the fragment size of our DNA. We were sequencing primarily shorter strands (less than 3.6kb) The optimal size is between 8-10kb. By trying a number of tricks, we began to ameliorate this problem.
DNA Isolation for Quality DNA:
Initially, we were using a bead beating protocol for isolation. This is a common tactic to break the tough cell walls of theorganism. Unfortunately, the resulting DNA samples had a considerable number of sheared fragments.
This was evident in both the gel electrophoresis analysis and the statistics of the sequencing run (see figure 1). To optimize this parameter, we decided to change our methods of isolation and purchased Thermo Fisher’s ‘Yeast DNA Extraction Kit’ (Product
#78870). This optimization dramatically improved our fragmentation problem and highlights the importance of high quality, high molecular weight DNA (for more information about the DNA input criteria for the MinION, look here). In conclusion, we suggest you use this kit if you are interested in yeast genome sequencing. Side Note: we have not gotten usable data from the MinION yet. Take all our advice with a grain of salt.
quality assembly (look here for more discussion on Quality score). We were having a difficult time achieving adequate quality scores (figure C3). Upon switching to the Thermo-Fisher Yeast Genomic isolation kit our quality scores dramatically improved (Figure D3). Through searching the Nanopore Community, we discovered a few key variables that contribute to quality score.
First, smaller fragments have a lower quality score (more information about Quality score here). The speed of translocation through the nanopore also contributes to the quality score. If the fragments move through the pore too fast, the quality score decreases. Furthermore, this translocation speed is dictated by the concentration of fuel, the intrinsic properties of the motor protein, and temperature (learn more about flow cell chemistry here). The fuel is contained within the Running Buffer and even 1uL extra of
Running Buffer can increase the translocation speed by 20%,
so make sure your pipettes are properly calibrated.
The flowcells have a sequencing life of 48 hours. However, a successful sequencing run does not require 48 hours to produce enough meaningful data, thus the flowcell can be re-used. Flowcell reuse comes with its own set of hurdles. To maintain the integrity of the flowcell, it must be washed using the Flowcell Wash Kit afterevery run, stored under proper conditions, and the sensor array must always be covered in buffer. In addition, voltage drift, the result of a depletion in the redox chemistry of the bulk solution is one of the major problems that arises with reuse. A voltage of
-180mV is sufficient to produce an adequate population of
active pores in a fresh flow cell. Over time, the common voltage needed to activate an adequate number of pores decreases (Here is a great video describing this phenomena).
Furthermore, optimization of the starting voltage is necessary to achieve the maximum number of pores available for sequencing and prolong the life of our flowcell. This change in common voltage is accomplished by adjusting the starting potential, based on the depicted scheme. The basic script from MinKNOW doesn’t allow variability in the starting potential, so users must manually edit the script to encode a pop-up box that allows changes to the starting potential (directions for editing the script can be found here.)
The fact remains that, with the combined effort of a Senior Research Scientist, a post-doc, and a graduate student, our lab has yet to produce adequate data from a single run to achieve our goal of de novo assembly of the yeast genome. None of these blog posts have discussed the bioinformatics aspect of the technology, which is an undertaking, to say the least. We still struggle with low performing flowcells that haven’t reached their expiration, low quality scores, an considerable fragmentation of our DNA sample.
We hope these optimization suggestions help you get past the learning curve a little faster and allow you to collect meaningful data with more ease. For more information on how the Oxford Nanopore’s MinION works and our critiques of the technology, see our first and second blog posts.
Looking for a instructions on how to load a prepped library onto the MinION flow cell? Check out a video our lab made.