Data-Efficient, Precise Calibration of Optical Tweezers Using Stochastic Force Inference

Precise quantification of microscopic forces in optical trapping is fundamental to a wide range of single-molecule biophysics. Conventional calibration techniques, however, require extensive datasets (typically exceeding 10⁴ samples) to achieve statistical convergence in the presence of thermal noise and nonequilibrium artifacts. In this work, we employ the stochastic force inference (SFI) method, which combines information theory with statistical physics, to reconstruct deterministic force fields from noisy stochastic trajectories in optical traps. We demonstrate that the SFI-based calibration offers exceptional advantages over traditional approaches, reducing the required acquisition duration by an order of magnitude while improving both accuracy and robustness. A femtonewton-level force spectral sensitivity of 14.84 fN Hz^{- 1 / 2} was achieved in solution, along with a minimum detectable force of 3.56 ± 0.96 fN, approaching the thermodynamic limit of optical force detection. We further applied the method to double-stranded DNA (dsDNA) stretching experiments across low- and high-force regimes, capturing signatures of both entropic bending rigidity and enthalpic stretching at a nanoscale resolution. Our results establish SFI as a data-efficient and noise-tolerant approach for optical tweezers calibration and molecular nanotechnology, with significant practical potential for biophysics and precision measurement.