Weight-Controllable Biobarcode Probes for Multi-input Breast Cancer Diagnosis.

Abstract

DNA-based molecular classifiers have emerged as a promising strategy for precise cancer diagnosis, offering a superior alternative to invasive biopsy detection. However, current DNA-computation-dependent molecular classifiers remain limited by complex pathways and cumbersome weight assignment procedures. To address this, we developed weight-controllable biobarcode probes (WBPs) that enable programmable signal amplification via precise stoichiometric regulation of barcode strands versus nonbarcode strands. These probes demonstrated robust performance in the weighted molecular computation. Using these WBPs, we performed fluorescence-based analog-to-digital signal conversion, enabling the representation of 128 combinations across up to seven targets. By integrating three miRNA inputs trained in silico machine learning, we constructed a WBP-based molecular classifier that can distinguish breast cancer patients from healthy individuals, achieving an accuracy of 84.00% on clinical serum samples. This work expands the scope of biobarcode technology from single-target detection to logical analysis of multiple targets, establishing a scalable and noninvasive platform for precision cancer diagnosis.