DNA-based Biomolecular Quantification and Molecular Manipulation
Name
wu-emwu-phd-meche-thesis-2026.pdf
Size
200.19 MB
Format
Adobe PDF
Checksum (MD5)
ebf4b41c96bf6d08f759d84a8f887ce1
Author(s)
Wu, Emily
Advisor(s)
Gopinath, Ashwin
Date Issued
February 2026
Publisher
Massachusetts Institute of Technology
Abstract
This thesis explores DNA-based methods of biomolecular quantification and molecular manipulation. Two axes of improvement in biomolecular quantification are investigated: (a) multiplexing, the simultaneous quantification of many targets, and (b) sensitivity, the quantification of low concentration targets. Challenges faced by current multiplexed quantification techniques include limited readout multiplexing and high cross-reactivity, while sensitive quantification techniques must balance sensitivity and assay complexity or time. Improvements along these axes can facilitate insight into biological mechanisms and drug development via highly multiplexed quantification of biomolecules, or potentially enable earlier diagnosis and intervention through the detection of low abundance disease biomarkers.
Perturbative quantification is an approach to highly multiplexed quantification that takes advantage of DNA-based chemical reaction networks (CRNs), advances in sequencing over the past several decades, and machine learning. In this approach, the equilibrium of a DNA-based CRN is perturbed by introducing a multiplexed sample to produce a unique multiplexed DNA signature. This signature can efficiently be read out via sequencing. With sufficient training data, a simple multilayer perceptron (MLP) network can be trained to predict an unknown sample composition given its associated DNA signature. We explore the theoretical foundation for this approach, and test its validity and capabilities by simulating systems that vary in size and complexity. We use the simulated data to train MLP models to accurately predict sample compositions of 20 different targets with high cross-reactivity. Thus we provide a scalable and flexible framework to simultaneously quantify multiple targets.
Next, we develop an approach to sensitive biomolecular quantification of targets with two distinct binding domains by forming a magnetic DNA origami conjugate via the capture of a single target of interest. Target capture is achieved in solution using a disc-shaped origami functionalized with a single primary binding moiety and 50 fluorescent dyes, a biotinylated secondary binding molecule, and streptavidin-functionalized iron oxide nanoparticles (IONPs). Captured targets are then rapidly immobilized by the application of a magnetic field onto a biotinylated surface passivated with dichlorodimethylsilane and Tween-20. The surface is then imaged via fluorescence microscopy to count individual targets. As an initial test of the approach, we quantify a single stranded DNA target in buffer, achieving a lower limit of quantification of 1.6 fM, with an operating range of 2.74 orders of magnitude. This method can easily be adapted to target antigens with two binding epitopes.
Lastly, we explore the use of the disc origami to control the immobilization of perovskite quantum dots (PQDs) on a planar semiconductor substrate, in an initial foray towards the goal of fabricating photonic integrated circuits for quantum information processing. Our approach is to take advantage of the ability to robustly control DNA origami immobilization on planar surfaces, combined with the excellent optical properties of CsPbBr3 PQDs as single photon emitters. We focused our efforts on three fundamental components: immobilization stability of origami under relevant processing conditions, DNA origami-PQD conjugation, and surface passivation to prevent the nonspecific immobilization of PQDs to the substrate. This lays the groundwork for the development of photonic integrated circuits for quantum information processing.
MIT Department
Massachusetts Institute of Technology. Department of Mechanical Engineering
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
