← All simulations

DNA Cyclization J Factors (Shimada–Yamakawa)

The J factor is the effective concentration of one end of a DNA molecule near the other, correctly aligned for ligation into a circle. For short DNA it depends strongly on length: bending a short piece into a circle costs energy, so J rises steeply to a maximum near 500 bp, and the ends must also be in register around the helix, so J oscillates with a period of one helical repeat. At long lengths the chain behaves like a Gaussian coil and J falls slowly as the ends become far apart. The curves are calculated from the theory of Shimada and Yamakawa (1984) for a twisted wormlike chain, from the persistence length, torsional modulus and helical repeat of the DNA. Use “Save current curve” to compare up to three parameter sets. Move the pointer over either plot to read off values.

Settings
nm
10−19 erg·cm
bp/turn
bp
bp
Advanced options
K. Enters only through the bending rigidity A = kBTP in Poisson’s ratio σ; the spreadsheet used 300 K.
in units of 2P (2.2 to 3.6 is about 615 to 1000 bp for P = 47.5 nm). The short-chain formula is used below this range and the long-chain formula above it; the end must stay below about 4, where the short-chain formula starts to diverge.
Curves on the plots
Saved curves (up to 3) remain when you change current settings.

J factor, log scale

log10 of J in mol/L

J factor, linear scale

J in nM

How this is calculated

The page uses the closed-form fitting functions of J. Shimada and H. Yamakawa (1984), “Ring-closure probabilities for twisted wormlike chains. Application to DNA,” Macromolecules 17, 689–698, as implemented in J. Kahn’s spreadsheet SY_calculations (equation numbers below are those of the paper, as noted in the spreadsheet). The DNA is a uniform, unbent, twisted wormlike chain with bending persistence length P, torsional modulus C, and helical repeat h, with a rise of 0.34 nm per base pair. Lengths L are in units of the Kuhn length 2P, and Poisson’s ratio σ couples bending and twisting (eqn 5):

sigma = A/C - 1, A = kB T P, L = N x 0.34 nm / 2P

Short and intermediate lengths. Closing the circle requires the two ends to meet, with parallel tangents and with the helix in register. The twist mismatch is the distance of the number of helical turns N/h from the nearest whole number, and the page adds the contributions of that mismatch and of the next whole turn either way:

Delta N = |N/h - round(N/h)|, delta in {Delta N, Delta N - 1, Delta N + 1}

For each, the ring-closure probability density is (eqns 37, 38 and 70)

G = C0 L^-6.5 exp[-(pi^2/L)(1 + delta^2/(1+sigma)) + (C1 + 1/4) L]
C0 polynomial
C1 polynomial
C1 coefficients

with C0 and C1 set to zero when |δ|/(1 + σ) > 1.38, where the fitting functions stop applying. The J factor in mol/L is then (eqns 72 and 73)

J = 8 pi^2 sum G / (NA (2P)^3)

The exponential term is why J falls so steeply below about 200 bp: the bending energy of a small circle grows as 1/L. The δ2 term is the twisting energy needed to bring the ends into register, which makes J oscillate with a period of one helical repeat, most strongly for short DNA.

Long lengths. The expansion above diverges for long chains (beyond about L = 4, 1100 bp for P = 47.5 nm), so for long DNA the page uses Shimada and Yamakawa’s weighting-function result for long chains:

J = (F0 + F1 cos(2 pi N/h)) / (NA (2P)^3)
F0 polynomial in L^-1/2
F1 = exp(-(2+sigma)L) polynomial in 1/L

The two expressions do not agree in the range where both might apply. The long-chain formula gives a mean J about 10–15% lower, and its twist oscillation is much smaller: F1 decays as e−(2+σ)L, so at 800 bp it gives oscillations of about ±9% where the short-chain formula gives about ±25%. Switching abruptly from one to the other therefore makes the oscillations collapse at one length. Instead, the page uses the short-chain formula below L = 2.2 (about 615 bp), the long-chain formula above L = 3.6 (about 1000 bp), and in between blends the two smoothly on a logarithmic scale, so the oscillations die away gradually. (The spreadsheet switched abruptly at about 780 bp.) The shaded band on the plots shows the blend range, which can be changed under Advanced options. F0 tends to the Gaussian-chain result at long lengths.

Gaussian chain. For a long chain, the ends meet at random orientations, and J is the probability density of zero end-to-end distance for a Gaussian coil with the mean-square end-to-end distance of a wormlike chain of contour length Lc (Jacobson and Stockmayer, 1950):

J_G = (3/(2 pi <r^2>))^(3/2) / NA, <r^2> = 2 P Lc - 2P^2 (1 - exp(-Lc/P))

It is shown as a dashed curve. It is far too high for short DNA, which is too stiff to be a coil, and the Shimada–Yamakawa result approaches it from below as the length increases.

Difference from the spreadsheet. In the long-chain formula the spreadsheet computes the twist term as cos(τ0L) with τ0 in Å−1 and L in units of 2P, which mixes units. The page uses cos(2πN/h), the twist phase of the whole molecule, which is τ0 times the contour length in Å. This changes J by up to about 15% around 800 bp and has no effect at long lengths, where F1 is negligible. Below the blend range the page reproduces the spreadsheet exactly.

Lengths are in base pairs (0.34 nm each), the persistence length in nm, C in units of 10−19 erg·cm, and J in mol/L (log plot) or nM (linear plot). Measured cyclization J factors depend on solution conditions, especially Mg2+, which changes the effective helical repeat and stiffness.

This page was checked by its author on desktop browsers in light mode. No human testing was done in dark mode or on phones.

Jason D. Kahn, Dept. of Chemistry and Biochemistry, University of Maryland College Park, using Claude

Table of values for the current curve

The table lists whole base-pair lengths across the current range. To use these numbers elsewhere, click Download .csv: the file opens directly in Excel, Numbers or Google Sheets, and in Matlab T = readtable('filename.csv') reads it into a table. Or click Copy table and paste into a spreadsheet.