INSTITUTE OF PHYSICS PUBLISHING
NANOTECHNOLOGY
Nanotechnology 17 (2006) 3160–3165
doi:10.1088/0957-4484/17/13/014
Electrical signatures of single-stranded
DNA with single base mutations in a
nanopore capacitor
Maria E Gracheva1 , Aleksei Aksimentiev1,2 and
Jean-Pierre Leburton1,3,4
1
Beckman Institute for Advanced Science and Technology, University of Illinois at
Urbana-Champaign, Urbana, IL 61801, USA
2
Department of Physics, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA
3
Department of Electrical and Computer Engineering, University of Illinois at
Urbana-Champaign, Urbana, IL 61801, USA
E-mail: [email protected] and [email protected]
Received 16 March 2006, in final form 2 May 2006
Published 5 June 2006
Online at stacks.iop.org/Nano/17/3160
Abstract
In this paper, we evaluate the magnitude of the electrical signals produced by
DNA translocation through a 1 nm diameter nanopore in a capacitor
membrane with a numerical multi-scale approach, and assess the possibility
of resolving individual nucleotides as well as their types in the absence of
conformational disorder. We show that the maximum recorded voltage
caused by the DNA translocation is about 35 mV, while the maximum voltage
signal due to the DNA backbone is about 30 mV, and the maximum voltage
of a DNA base is about 8 mV. Signals from individual nucleotides can be
identified in the recorded voltage traces, suggesting a 1 nm diameter pore in a
capacitor can be used to accurately count the number of nucleotides in a
DNA strand. Furthermore, we study the effect of a single base substitution on
the voltage trace, and calculate the differences among the voltage traces due
to a single base mutation for the sequences C3 AC7 , C3 CC7 , C3 GC7 and
C3 TC7 . The calculated voltage differences are in the 5–10 mV range. The
calculated maximum voltage caused by the translocation of individual bases
varies from 2 to 9 mV, which is experimentally detectable.
(Some figures in this article are in colour only in the electronic version)
In the early 1990s, it was suggested that the chemical structure
of nucleic acids can be resolved using natural nano-scale
features of a membrane channel [1]. Indeed, when an electric
field drives a DNA molecule through a biological channel
suspended in a lipid bilayer, the molecule partially blocks the
pore which is reflected by a transient reduction of the ionic
current; the duration of the current blockade is proportional
to the DNA length [1–7]. Statistical analysis of many current
blockades allowed the researchers to discriminate between
different sequences of RNA [3] and DNA [4] homopolymers,
blocks of purine or pyrimidine nucleotides within single
RNA [3] and DNA [8] strands, and, recently, the global
orientation of the nucleic acids in the pore [9, 10]. The
4 Author to whom any correspondence should be addressed.
0957-4484/06/133160+06$30.00
ionic current blockades were found to be sensitive to a single
nucleotide substitution in DNA and RNA hairpins [5, 11, 12],
raising the prospect of creating a nanopore sensor capable of
reading the nucleotide sequence directly from a DNA or RNA
strand.
A robust alternative to biological membrane channels
are artificial nanometre-sized pores in solid state membranes
manufactured by using highly focused electron or ion
beams [13–17]. As in the case of proteinaceous nanopores,
the correlation between DNA translocation and ionic current
blockage was established [17, 19–23].
In addition to
ionic current blockage, translocation of DNA molecules
through nanopores in metal–oxide–semiconductor membrane
is accompanied by a change in the electrostatic potential on
© 2006 IOP Publishing Ltd Printed in the UK
3160
Electrical signatures of single-stranded DNA with single base mutations in a nanopore capacitor
Figure 1. (A) DNA strand enveloped by a cylindrical surface, a phantom pore, initially wide enough to accommodate the strand entirely.
(B) The conformation of a DNA strand in a 1 nm diameter pore. The DNA backbone is stretched, and the DNA bases are tilted towards the
5 end of the strand. (C) Contour plots of negative charge concentration in the capacitor membrane and electrolyte solution. (D) Same for the
positive charge concentration.
the semiconductor layers of the membrane. Hence, as DNA
permeates the pore, its electronic signature can be recorded in
the form of a voltage trace.
Sequencing DNA strands with a nanopore device has
not yet been achieved. Although it has been demonstrated
that voltage signals resulting from DNA translocation through
a capacitor membrane can be recorded [24, 25], the
temporal resolution of such measurements precludes us from
recording signals that could be associated with translocation
of individual nucleotides. While new experimental methods
for measuring fine signals with nanopore sensors are
being developed [24, 27–30], the theoretical resolution of
such measurements can be estimated through computation.
Recently, we developed a multi-scale method for simulation of
the electrical response produced by DNA translocation through
a nanopore in a semiconductor capacitor membrane [25]. In
this approach, the DNA translocation through the nanopore is
simulated by molecular dynamics, while the voltage response
in the solid-state capacitor is modelled by a 3D self-consistent
Poisson Solver (PS).
In this paper, we calculate the electrical signals produced
by DNA translocation through a capacitor membrane with our
multi-scale approach, and assess the possibility of resolving
individual nucleotides as well as their types.
Ultra small pores in capacitor membranes are manufactured by exposing the membranes to a focused, high-energy
electron beam [16, 24]. By varying the focus, the energy and
the exposure time to the beam, pores of different diameters
can be manufactured. As the conformation of DNA in a pore
depends on the pore diameter, the latter will also affect the
strength of the recorded signals. When a single DNA strand is
confined to a pore smaller than 1.5 nm in diameter, the bases of
DNA tilt collectively towards the 5 end of the strand [9]. In a
1 nm diameter pore, the DNA bases, as well as their sequence
specific dipole moments [26], orient along the pore axis; the
fluctuations of the DNA conformation are greatly suppressed;
both factors contribute to a strong sequence-dependent signal. It has been demonstrated experimentally that single DNA
strands can permeate a 1 nm diameter pore [31], which we consider in this study.
Since A, C, G, and T DNA nucleotides differ from each
other by only a few atoms, it is essential to characterize the
conformation of DNA inside the pore in atomic detail in order
to relate the DNA sequence to the measured electrical signals.
In this study, we rely on molecular dynamics (MD) [32]
to obtain realistic snapshots of DNA conformations in the
nanopore. The AMBER95 force field [33] accurately describes
DNA dynamics in solution [34, 35], as well as the distribution
of the electrostatic potential around DNA [36, 37], which
can be regarded as the electrostatic fingerprint of the DNA
sequence. The interaction of the DNA charge with the
charge distribution in the semiconductor layers of the capacitor
membrane is taken into account by a 3D self-consistent PS
scheme, that also accounts for the screening of the DNA charge
by counterions in the solution, and for the redistribution of
electrons and holes in the semiconductor materials and the
oxide [25].
As the structure of the silicon oxide surface in the pore
is not known to atomic detail yet, we carried out our MD
simulations using a phantom pore model [9, 31]. A DNA strand
comprised of eleven nucleotides was placed in a rectangular
volume of pre-equilibrated TIP3P water molecules; K+ and
Cl− ions were added, corresponding to a 1 M concentration.
The DNA strand was enveloped by a cylindrical surface, a
so-called phantom pore, that was initially wide enough to
accommodate the strand entirely (figure 1(A)). In a 1 ns
MD simulation, the surface was gradually shrunk to a 1 nm
diameter while 10 pN forces pushed DNA atoms lying outside
of the surface towards the centre of the pore. At the end of the
simulation, the DNA strand adopted a stretched conformation
in which all DNA bases were tilted towards the 5 end of the
strand (figure 1(B)).
In order to estimate the energy costs associated with
DNA confinement, we computed the conformational energy
of the confined and unconfined DNA oligomers: 620 ± 26
3161
M E Gracheva et al
(constrained) and 536 ± 17 (unconstrained) kcal/mol for
C3 AC7 . We found that the conformational energy increases
significantly when DNA oligomers are confined to a 1 nm
diameter pore. However, the energy increase per nucleotide
(ranging from 2.7 to 11.5 kcal mol−1 ) is still smaller than the
gain in the electrostatic energy associated with translocation of
one nucleotide (total charge—e) across the membrane at a 1 V
bias (23 kcal mol−1 ). One should also take into account the
fact that in a real pore, which has an hourglass shape, only a
3-nucleotide fragment of DNA is subject to the confinement,
whereas the entire strand experiences the electrostatic force
driving it through the pore. The above estimates are rather
crude, as we did not take into account the interactions of
DNA oligomers with their environment: water, ions and the
nanopore. Such evaluations are complicated by the fact that
the morphology and the physical state of the nanopore surface
are not known. Recent experiments with synthetic [17, 31] and
biological [1, 3, 4] nanopores have unequivocally demonstrated
that single DNA strands can permeate pores that are only
1.0–1.3 nm in diameters. Therefore, we believe that the
conformation of the DNA used in our studies is energetically
feasible.
Although it is possible to simulate the entire process
of DNA translocation with our multi-scale method, in this
study we eliminate, for the time being, the effects of the
conformational anisotropy on the magnitude of the recorded
signals, as well as the temporal variation of the signal due
to DNA translocation, focusing entirely on the profile of
the voltage signals as a function of the DNA location in a
nanopore. For this purpose, we first determine a representative
conformation of DNA in a nanopore, and then, using a
simple translation, emulate DNA translocation through the
pore, keeping the DNA conformation constant.
In a real pore, only part of the DNA is confined to
the fullest extent, while away from the pore constriction the
DNA is not restrained. We have already carried out MD
simulations of DNA translocation through a 1 nm diameter
pore and observed the compression of DNA as it passed
through the narrowest part of the pore followed by the rebound
to the unrestrained conformation once DNA escapes the pores
constriction [31]. However, in this study we limit ourselves
to the analysis of a system in which we have control over
the conformation of DNA in the pore, eliminating for now
from our consideration the effect of conformational anisotropy,
which might be adjustable in experiment. As we demonstrate
below the electric signal from a DNA base spreads over its
next neighbours only. Therefore, the effect of the DNA
conformation outside the pore on the voltage trace recorded
at the junction of silicon and silicon dioxide is negligible.
The modelled capacitor membrane consists of a 2 nm
thick SiO2 layer sandwiched between two heavily doped n+ –Si
layers of 4.5 nm thickness each. The nanopore in the capacitor
membrane has a double-conical shape with the diameter of
the narrowest part being 1 nm (see details in [25]). Taking
the distribution of the DNA’s charge from MD simulations,
we solve self-consistently the Poisson equation, assuming
the ions in the electrolyte obey the Boltzmann distribution,
whereas electrons and holes in the semiconductor are ruled
by the Fermi–Dirac statistics. This model was described
elsewhere in detail [25]. The average ion concentrations that
3162
DNA front position, A
DNA front position, A
Figure 2. (a) Calculated voltage trace from C3 AC7 strand
(circle-line) and the backbone signal of the same strand (solid line).
(b) The difference between the whole DNA signal and the backbone
signal.
we obtain in the pore correlate with the data obtained by
molecular dynamics simulations. Since we are interested in
the average induced signal on the electrodes, we believe that
our solution model accurately describes the solution dynamics.
The concentration of the KCl electrolyte is 1 M, the doping
density in silicon is Nd+ = 2 × 1020 cm−3 . We also include in
our model a thin (2 Å) layer of negative charge at the surface
of the semiconductor structure with the density of 1021 cm−3 ,
resulting from the etching process [18].
The calculated contour plots of negative and positive
charge concentration in the capacitor membrane and electrolyte
solution are shown in figures 1(C) and (D), respectfully, for one
of the DNA positions in the pore. The cross-section was taken
through the middle of Si–SiO2 –Si layers and the centre of the
nanopore. It can be seen that the central SiO2 layer is void of
charge carriers. There is a distribution of electrons in the Si
layers with a depletion region near the surface due to the fixed
surface charge (figure 1(C)). The accumulation of the positive
charge is present around the DNA as it is negatively charged
(figure 1(D)).
While DNA is moved in the fixed conformation along
the axis of the nanopore in 1 Å increments, the electrostatic
potential induced by the DNA charge is recorded on the lower
and upper Si electrode as a function of the DNA position in
the nanopore (see [25] for full description). The DNA is at
‘zero’ position when the front tip of the molecule is at the
nanopore centre. As the signals obtained from the upper and
lower electrodes do not differ significantly, we focus on the
voltage traces calculated on the upper Si electrode, similar to a
different single electrode approach [27].
In figure 2(a), the recorded voltage trace for an eleven
base DNA with the sequence 5 -CCCACCCCCCC-3 is shown
in blue. To simplify the notation, we abbreviate this DNA
sequence as C3 AC7 . The other studied DNA sequences
are abbreviated in a similar way. The voltage trace of the
DNA’s backbone which was obtained by removing the bases
Electrical signatures of single-stranded DNA with single base mutations in a nanopore capacitor
C3-A-C7
0
C3-C-C7
-10
C3-G-C7
C3-T-C7
-20
-40
-20
0
20
DNA front position, A
40
-30
60
-40
is shown in red on the same figure. The difference between
these two signals is displayed in figure 2(b). From the two
signals shown in figure 2(a) we conclude that the maximum
recorded voltage caused by the DNA translocation is around
35 mV. This signal is appreciable and is due to the DNA
confinement in the constriction where screening by the solution
is reduced. In addition, due to the narrowness of the pore the
DNA translocates in close proximity to the electrodes. The
maximum voltage induced by the DNA’s backbone is about
30 mV, and therefore the maximum signal from a base is
approximately 8 mV (figure 2(b)). We point out that these
numbers are obtained in the absence of the conformational
disorder. Also, measured from the non-zero voltage trace, the
DNA’s length is ∼70 Å, which is in agreement with the actual
stretched DNA length. There are several distinct fluctuations
on the voltage trace, which, as we show later, correspond to
the 11 segments of the backbone with a DNA base in every
segment.
In order to better understand the observed variations in the
voltage trace we separate the C3 AC7 DNA into 11 fragments,
each containing a sugar ring, a phosphate group and a base
(with the exception of the first fragment which does not contain
the phosphate group). Each fragment was driven through
the pore separately and the corresponding voltage signal was
calculated at the electrode. The composite plot of traces of the
eleven fragments and the trace of the whole C3 AC7 sequence
is shown in figure 3. We observe that 10 of the 11 fragments
of the DNA produce similar traces in the form of a negative
dip. The remaining one (trace #1 on figure 3) corresponds
to the first fragment of the DNA with no phosphate group in
it. The absence of the phosphate group, which is negatively
charged, allows for a positive rise in this trace caused by the
positive charge of the sugar ring and the base in this segment.
We also observe that the traces of the DNA fragments overlap
significantly, as the average width of each dip extends over 8–
12 Å. This means that in the whole DNA trace the signals from
as many as three nearby fragments (and therefore bases) are
mixed, which is to be expected as the Coulomb interaction is
a long range interaction. In addition, trace #4, which contains
the ‘A’ base does not have any obvious distinct features from
other segments containing the ‘C’ bases, except maybe a slight
widening of the signal compared to the nearest fragments,
which may be due to the different distribution of charge on
the base as well as to the different orientation of the base #4 on
the backbone, or, most likely, both.
-20
0
20
40
60
DNA front position, A
Figure 4. Calculated voltage traces for four DNA strands C3 AC7 ,
C3 CC7 , C3 TC7 and C3 GC7 obtained by a single base mutation at the
fourth nucleotide.
Voltage, mV
Figure 3. Voltage traces of the eleven single-nucleotide fragmets and
the signal of the whole C3 AC7 DNA sequence. The contribution
from individual nucleotides can be seen in the total trace, and, thus
one can count the number of nucleotides in the strand.
DNA front position, A
Figure 5. Voltage traces due to translocation of single bases C, A, T
and G. The inset illustrates the conformations of the replaced
nucleotides; the orientation of their aromatic rings as well as the
backbone structure are identical.
To study the effect of single nucleotide substitution on
the signal induced on the capacitor structure, we replace one
base (#4) in the C3 AC7 strand by C, G, and T nucleotides
respectively, using the psfgen building module of NAMD [38].
We keep the conformations of the three resulting strands
identical to that of the original strand, with the exception of
the replaced nucleotides. The resulting DNA strands have
sequences C3 CC7 , C3 GC7 and C3 TC7 , correspondingly. The
inset in figure 6 illustrates the conformations of the replaced
nucleotides; the orientation of their aromatic rings as well as
the backbone structure are identical. The calculated voltage
traces for the four strands obtained by a single base mutation
at the fourth place are illustrated in figure 4. It is clear that the
signals differ in the vicinity of the fourth base over a position
range between −15 and 5 Å. As expected, the traces are
identical away from the fourth base as the DNA conformations
there remained the same.
In order to understand the differences in obtained DNA
traces, we consider a purely academic case with translocation
of single C, A, T and G bases without an attached backbone
fragment (figure 5). The shape of the voltage traces for
C, A and G is similar to the trace one would obtain for
3163
Voltage, mV
Voltage, mV
M E Gracheva et al
C3TC7(-)C3CC7
C3TC7(-)C3AC7
DNA front position, A
Voltage, mV
Voltage, mV
DNA front position, A
C3TC7(-)C3GC7
C3AC7(-)C3CC7
DNA front position, A
Voltage, mV
Voltage, mV
DNA front position, A
C3AC7(-)C3GC7
DNA front position, A
C3CC7(-)C3GC7
DNA front position, A
Figure 6. (a) Signal differences between single bases T and A, as well as C3 TC7 and C3 AC7 ; (b) differences between single bases T and C, as
well as C3 TC7 and C3 CC7 ; (c) differences between single bases A and C, as well as C3 AC7 and C3 CC7 ; (d) differences between single bases T
and G, as well as C3 TC7 and C3 GC7 ; (e) differences between single bases A and G, as well as C3 AC7 and C3 GC7 ; (f) differences between
single bases C and G, as well as C3 CC7 and C3 GC7 .
the translocation of a dipole (positive and negative charges
separated by a fixed distance) [25]. The signal for the T
base is somewhat more complex and could be interpreted, for
example, as the translocation of a sequence of three charges
(negative–positive–negative). Note, that the orientation of the
bases in the pore is similar and fixed, while, in principle,
the recorded shape of the signals depends on the orientation
and position of the base in the pore. The maximum induced
voltage variation is obtained for the C base (9 mV), whereas
the minimum variation is recorded for the T base (2 mV). Also,
four curves form two groups: C and T traces have similar wider
signals, whereas A and G traces have similar narrower signals.
This correlates with the molecular structure of the bases: C and
T have similar structure as pyrimidine bases, whereas A and G
have similar structure as purine bases.
We analyse the traces in figures 4 and 5 with respect
to each other by subtracting them in corresponding pairs in
order to obtain an estimate of the voltage difference between
two traces with a single base mutation in the absence of
conformational disorder.
The results are shown in figures 6(a)–(f). Figure 6(a)
shows two trace differences: the first difference is the
difference between traces caused by the translocation of single
bases T and A, the second is the difference between traces
caused by the translocation of C3 TC7 and C3 AC7 . It can
be seen that the differences have essentially the same shape,
which suggests that the signal difference between C3 TC7 and
3164
C3 AC7 comes, mostly, from the base substitution. Some
deviation of the curves is expected as the bases are screened
by the backbone in one case and are not in the other. Also, the
screening by the solution is different in both cases, especially
away form the centre of the bases. These features are
representative of all pairs of curves in figures 6(a)–(f). Hence,
the difference in the recorded DNA traces with a single base
mutation comes mostly from the base substitution. One cannot
emphasize enough the importance of these results as they
become accessible to experiment and could be the basis for
discriminating between different DNA bases, leading to the
identification of the whole DNA molecular sequence.
In conclusion, we studied the electrical signals produced
on a capacitor membrane containing a 1 nm nanopore by single
stranded DNAs with a single base mutation. The calculated
maximum voltage due to the different bases varies from 2
to 9 mV, which is experimentally detectable. Signals from
individual nucleotides can be identified in the recorded voltage
traces, suggesting a 1 nm diameter pore in a capacitor can
be used to accurately count the number of nucleotides in a
DNA strand. Our data show the possibility of distinguishing
DNAs with different sequences in the ideal situation when the
strands have the same conformation. However, the influence
of stochastic DNA translocation on the induced signal is an
important issue for realistic modelling of the nanopore—DNA
system. The stochastic ‘wiggling’ of the DNA, although
restricted by the nanopore, may reduce the resolution of single
Electrical signatures of single-stranded DNA with single base mutations in a nanopore capacitor
nucleotides or even make nucleotides indistinguishable. This
issue is currently under investigation and will be the subject
of a subsequent publication. Nevertheless, our results are
an important milestone in DNA detection because techniques
could be used to reduce or annihilate DNA stochastic
movement in a nanopore. One such recently proposed
technique involves wrapping DNA on a carbon nanotube [27],
which is then lowered into a pore to detect bases. Another
approach for reducing the effect of the DNA stochastic motion
on the recorded voltage, is the method of ‘DNA flossing’
through a nanopore to sample segments of the molecule and
collect the average reading [39].
Acknowledgments
This work was funded by NIRT-NSF grant CCR0210843, Darpa grant 392FA9550-04-1-0214, NIH grants ROIHG003713 and P41-PR05969.
The authors gratefully
acknowledge the use of the supercomputer time at the National
Center for Supercomputer Applications and the Pittsburgh
Supercomputer Center provided through Large Resource
Allocation Committee grant MCA05S028. We are grateful to
Gregory Timp and Klaus Schulten for useful discussions.
References
[1] Kasianowicz J J, Brandin E, Branton D and Deamer D W 1996
Proc. Natl Acad. Sci. USA 93 13770–3
[2] Song L, Hobaugh M R, Shustak C, Cheley S, Bayley H and
Gouaux J E 1996 Science 274 1859–66
[3] Akenson M, Branton D, Kasianowicz J J, Brandin E and
Deamer D W 1999 Biophys. J. 77 3227–33
[4] Meller A, Nivon L, Brandin E, Golovchenko J and
Branton D 2000 Proc. Natl Acad. Sci. USA 97 1079–84
[5] Vercoutere W, Winters-Hilt S, Olsen H, Deamer D,
Haussler D and Akeson M 2001 Nat. Biotechnol. 19 248–52
[6] Meller A, Nivon L and Branton D 2001 Phys. Rev. Lett.
86 3435–8
[7] Meller A and Branton D 2002 Electrophoresis 23 2583–91
[8] Wang H, Dunning J E, Huang A P H, Nyamwanda J A and
Branton D 2004 Proc. Natl Acad. Sci. USA 101 13472–7
[9] Mathé J, Aksimentiev A, Nelson D R, Schulten K and
Meller A 2005 Proc. Natl Acad. Sci. USA 102 12377–82
[10] Butler T Z, Gundlach J H and Trall M A 2006 Biophys. J.
90 190–9
[11] Howorka S, Cheley S and Bayley H 2001 Nat. Biotechnol.
19 636–9
[12] Ashkenasy N, Sánchez-Quesada J, Bayley H and
Ghadiri M R 2005 Angew. Chem. Int. Edn Engl. 44 1401–4
[13] Li J, Stein D, McMullan C, Branton D, Aziz M J and
Golovchenko J A 2001 Nature 412 166–9
[14] Li J, Gershow M, Stein D, Brandin E and
Golovchenko J A 2003 Nat. Mater. 2 611–5
[15] Storm A J, Chen J H, Ling X S, Zandbergen H W and
Dekker C 2003 Nat. Mater. 2 537–40
[16] Heng J B et al 2003 IEDM Tech. Dig. 767–70
[17] Heng J B, Ho C, Kim T, Timp R, Aksimentiev A,
Grinkova Y V, Sligar S, Schulten K and Timp G 2004
Biophys. J. 87 2905–11
[18] Ho C, Qiao R, Heng J B, Chatterjee A, Timp R J, Aluru N R
and Timp G 2005 Proc. Natl Acad. Sci. USA 102 10445–50
[19] Aksimentiev A, Heng J B, Timp G and Schulten K 2004
Biophys. J. 87 2086–97
[20] Mara A, Siwy Z, Trautmann C, Wan J and Kamme F 2004
Nano Lett. 4 497–501
[21] Chang H, Kosari F, Andreadakis G, Alam M A,
Vasmatzis G and Bashir R 2004 Nano Lett. 4 1551–6
[22] Storm A J, Chen J H, Zandbergen H W and Dekker C 2005
Phys. Rev. E 71 051903–13
[23] Fologea D, Gershow M, Ledden B, McNabb D S,
Golovchenko J A and Li J 2005 Nano Lett. 5 1905–9
[24] Heng J B, Aksimentiev A, Ho C, Dimitrov V, Sorsch T,
Miner J, Mansfield W, Schulten K and Timp G 2005 Bell
Labs Tech. J. 10 5–22
[25] Gracheva M E, Xiong A, Aksimentiev A, Schulten K,
Timp G and Leburton J-P 2006 Nanotechnology 17 622–33
[26] Saenger W (ed) 1984 Principles of Nucleic Acid Structure
(New York: Springer)
[27] King G M and Golovchenko J A 2005 Phys. Rev. Lett.
95 216103-d
[28] Fan R, Karnik R, Yue M, Li D, Majumdar A and Yang P 2005
Nano Lett. 5 1633–7
[29] Fologea D, Uplinger J, Thomas B, McNabb D S and Li J 2005
Nano Lett. 5 1734–7
[30] Zwolak M and Ventra M D 2005 Nano Lett. 5 421–4
Lagerqvist J, Zwolak M and Di Ventra M 2006 Preprint
arXiv:cond-mat/0601394
[31] Heng J B, Aksimentiev A, Ho C, Marks P, Grinkova Y V,
Sligar S, Schulten K and Timp G 2006 Biophys. J.
90 1098–106
[32] Allen M P and Tildesley D J 1987 Computer Simulation of
Liquids (New York: Oxford University Press)
[33] Cornell W D, Cieplak P, Bayly C I, Gould I R, Merz K M Jr,
Ferguson D M, Spellmeyer D C, Fox T, Caldwell J W and
Kollman P A 1995 J. Am. Chem. Soc. 117 5179–97
[34] Feig M and Pettitt B M 1998 Structure 6 1351–4
[35] Beveridge D L et al 2004 Biophys. J. 87 3799–813
[36] Schmidt M W et al 1993 J. Comput. Chem. 14 1347–63
[37] Bayly C I, Cieplak P, Cornell W D and Kollman P A 1993
J. Phys. Chem. 97 10268–80
[38] Phillips J C, Braun R, Wang W, Gumbart J, Tajkhorshid E,
Villa E, Chipot C, Skeel R D, Kale L and Schulten K 2005
J. Comput. Chem. 26 1781–802
[39] Kasianowicz J J 2004 Nat. Mater. 3 355–6
3165