Computational and molecular design of programmable oligonucleotide probes for multiplexed in situ detection

Abstract

Fluorescence in situ hybridization (FISH) has detected nucleic acid targets in biological samples for more than four decades, from mapping genes on chromosomes to spatially profiling thousands of RNA species in single cells. It exploits a simple physical principle: complementary nucleic-acid sequences preferentially hybridize to one another. Modern probes are computationally designed oligonucleotides carrying a sequence that recognizes the target together with auxiliary sequences that recruit signal, carry barcodes and prime amplification, each intended to bind only to its target sequence while avoiding everything else. This dissertation addresses the computational design and the molecular construction of programmable oligonucleotide probes for multiplexed in situ detection. PaintSHOP makes genome-scale probe design accessible without writing code, pairing an interactive web application with hosted, genome-wide probe collections and the pipeline that generates them. OligoMiner2 improves the computation underneath that interface, in its bioinformatics and in its models of duplex stability, raising the speed and fidelity of specificity scoring. Applied to a previously published probe library, it predicts off-target binding for 1,158 probes scored as carrying none. The same tools apply to synthetic sequences, characterizing a published 240,000-member orthogonal barcode set that satisfies sequence orthogonality criteria but still cross-hybridizes widely, and rebuilding it at matched composition for 16.7-fold fewer cross-hybridizing pairs. Splint Ligation Extension (SLX) is established as a modular probe platform to address molecular construction. A probe is assembled from a kit of oligonucleotide blocks joined at programmed junctions by splint-directed ligation, so the number of imager-binding sites it carries is specified at design time rather than acquired by stochastic elongation. The kit assembles programmed products in vitro and in multiplexed one-pot reactions, and those products are functionally validated on human metaphase chromosome spreads, in both a genomic repeat experiment and a multiplexed single-copy probe library. On a synthetic construct carrying six addressable readout domains, the number of readout domains targeted modulates the signal intensity in situ, while the signal from a reference chemistry in the same cells does not. PaintSHOP and OligoMiner2 address the requirements of probe design computationally, before a sequence is ordered; SLX addresses them molecularly, at assembly. Together they increase the fidelity of multiplexed FISH probe reagents and facilitate advanced multiplexing strategies, without the fluidic exchange that sequential barcoding requires. Reagents that behave as designed are what will bring spatial biology methods into the clinical settings that FISH has served for decades.