Improving sequencing by tunneling with multiplexing and cross-correlations

Abstract

Sequencing by tunneling is a next-generation approach to read single-base information using electronic tunneling transverse to the single-stranded DNA (ssDNA) backbone while the latter is translocated through a narrow channel. The original idea considered a single pair of electrodes to read out the current and distinguish the bases [1,2]. Here, we propose an improvement to the original sequencing by tunneling method, in which N pairs of electrodes are built in series along a synthetic nanochannel. While the ssDNA is forced through the channel using a longitudinal field it passes by each pair of electrodes for long enough time to gather a minimum of m tunneling current measurements, where m is determined by the level of sequencing error desired. Each current time series for each nucleobase is then cross-correlated together, from which the DNA bases can be distinguished. We show using random sampling of data from classical molecular dynamics, that indeed the sequencing error is significantly reduced as the number of pairs of electrodes, N , increases. Compared to the sequencing ability of a single pair of electrodes, cross-correlating N pairs of electrodes is exponentially better due to the approximate log-normal na- ture of the tunneling current probability distributions. We have also used the Fenton-Wilkinson approximation to analytically describe the mean and variance of the cross-correlations that are used to distinguish the DNA bases. The method we suggest is particularly useful when the measurement bandwidth is limited, allowing a smaller electrode gap residence time while still promising to consistently identify the DNA bases correctly.

Classification of DNA nucleotides with transverse tunneling currents

“…The maximum likelihood framework for nucleotide assignment is easily generalized to more complicated setups than just a single pair of electrodes (see, e.g., the setup in [29]), or extended to include other types of information than just the measured current values. Other aspects that could help the identification could be, e.g., the duration of current spikes, the time interval between spikes, and the fluctuations of currents within spikes [30].…”

“…Recently, it was investigated theoretically by simulations whether the use of multiple electrode pairs coupled in series could improve identification of nucleotides [29, 42]. The advantage of multiple electrodes is an increased number of measurements for each nucleotide and, consequently, a lower error rate.…”

“…Figure 2 shows an example of a current trace for the nucleotide A , and histograms of the current values for all four nucleotides are shown in figure 3 as obtained in reference [29]. We here plot the log-current probability distributions p ( Ĩ | X ) with Ĩ = log 10 ( I /Amp), and where X ∈ { A, T, G, C } denotes the four types of nucleotides.…”

“…Histograms of the probability distributions p ( Ĩ | X ) for the log-current values Ĩ = log 10 ( I /Amp) for the four different nucleotides (same as figure 2 in reference [29]). Dashed vertical lines mark points on the current axis where one distribution replaces another at being the one with the highest probability density.…”

“…Consequently, narrowly defined current values can be measured only with integration times much longer than microseconds. This limits the time resolution of current measurements, which can be ameliorated by multiplexing with several pairs of electrodes [29]. …”