BOCS: DNA k-mer content and scoring for rapid genetic biomarker identification at low coverage

A single, inexpensive diagnostic test capable of rapidly identifying a wide range of genetic biomarkers would prove invaluable in precision medicine. Previous work has demonstrated the potential for high-throughput, label-free detection of A-G-C-T content in DNA k-mers, providing an alternative to single-letter sequencing while also having inherent lossy data compression and massively parallel data acquisition. Here, we apply a new bioinformatics algorithm - block optical content scoring (BOCS) - capable of using the high-throughput content k-mers for rapid, broad-spectrum identification of genetic biomarkers. BOCS uses content-based sequence alignment for probabilistic mapping of k-mer contents to gene sequences within a biomarker database, resulting in a probability ranking of genes on a content score. Simulations of the BOCS algorithm reveal high accuracy for identification of single antibiotic resistance genes, even in the presence of significant sequencing errors (100% accuracy for no sequencing errors, and > 90% accuracy for sequencing errors at 20%), and at well below full coverage of the genes. Simulations for detecting multiple resistance genes within a methicillin-resistant Staphylococcus aureus (MRSA) strain showed 100% accuracy at an average gene coverage of merely 0.515, when the k-mer lengths were variable and with 4% sequencing error within the k-mer blocks. Extension of BOCS to cancer and other genetic diseases met or exceeded the results for resistance genes. Combined with a high-throughput content-based sequencing technique, the BOCS algorithm potentiates a test capable of rapid diagnosis and profiling of genetic biomarkers ranging from antibiotic resistance to cancer and other genetic diseases.