Profiling DNA mutation patterns for cancer classification plays an essential role in precision and personalized medicine. Conventional PCR-based mutation assay is limited by the extensive labour on target amplification. We herein create an amplification-free surface enhanced Raman spectroscopy (SERS) biochip which enables direct and simultaneous identification of multiple point mutations in tumor cells. Without pre-amplifying the target sequences, the SERS assay reads out the presence of cellular mutations through the interpretation of Raman fingerprints. The SERS sensor is integrated into a microfluidic chip, achieving one-step multiplex analysis within 40 min. Importantly, by combining SERS spectra encoding technique with supervised learning algorithm, a panel of nucleotide mixtures can be well distinguished according to their mutation profiles. We initially demonstrate an excellent levels of classification in samples from colorectal cancer and melanoma cell lines. For final clinical validation, the system performance is verified by classifying cancer patient samples, which shows an accuracy above 90%. Due to the simplicity and rapidness, the SERS biosensor is expected to become a promising tool for clinical point-of-care diagnosis towards precision medicine.

用于癌症分类的DNA突变图谱分析在精准和个性化医学中起着至关重要的作用。常规的基于PCR的突变分析受到靶标扩增方面大量工作的限制。我们在此创建了一种无扩增的表面增强拉曼光谱（SERS）生物芯片，该芯片能够直接并同时识别肿瘤细胞中的多点突变。在不预先扩增靶序列的情况下，SERS分析通过解释拉曼指纹来读出细胞突变的存在。SERS传感器集成到微流控芯片中，可在40分钟内完成一步式多重分析。重要的是，通过将SERS光谱编码技术与监督学习算法相结合，可以根据其突变谱很好地区分一组核苷酸混合物。我们最初在来自结肠直肠癌和黑色素瘤细胞系的样品中显示出极好的分类水平。对于最终的临床验证，通过对癌症患者样品进行分类来验证系统性能，其准确性可达到90％以上。由于其简单性和快速性，SERS生物传感器有望成为有希望的工具，可用于精确医学的临床即时诊断。