ULTRA-SENSITIVE LIQUID BIOPSY THROUGH DEEP LEARNING EMPOWERED WHOLE GENOME SEQUENCING OF PLASMA

Smart overview of the Invention

The patent application introduces a system leveraging deep learning for ultra-sensitive liquid biopsy through whole genome sequencing of plasma. This approach is focused on identifying circulating tumor DNA (ctDNA) in a patient's blood sample, which is indicative of cancer presence. The system utilizes advanced computational methods to classify and label sequence fragments that represent tumor markers, enhancing the early diagnosis and monitoring of cancer.

Technical Field

This innovation is situated within the realm of medical diagnostics, specifically targeting the detection of circulating tumor DNA for cancer diagnosis. It aims to improve the identification of low-abundance disease markers using sophisticated algorithms that can process complex genomic data obtained from non-invasive liquid biopsies.

Background

Cancer diagnosis traditionally relies on imaging techniques and tissue biopsies, which have limitations such as high false-positive rates and invasiveness. Liquid biopsies offer a non-invasive alternative by analyzing ctDNA from blood samples. However, existing methods struggle with sensitivity due to the low concentration of ctDNA amidst abundant cell-free DNA (cfDNA). The application addresses these challenges by employing deep learning models to enhance the detection capabilities in low tumor burden settings.

Methodology

The proposed method involves reading a plurality of reference sequences alongside sequence fragments from a patient's biological sample. Two reads are selected from these fragments, and a tensor is generated comprising reference sequences and positional data. A first trained classifier provides a regional probability based on patient features, while a second classifier with a convolutional neural network assesses local probabilities using the tensor. Tumor markers are labeled when both probabilities surpass predefined thresholds, indicating the presence of ctDNA.

Significance

By improving sensitivity and accuracy in detecting ctDNA, this system has the potential to transform cancer diagnostics. It enables early-stage cancer screening and monitoring with minimal invasiveness, offering clinicians valuable insights for managing disease progression and treatment outcomes. This advancement could lead to better survival rates and quality of life for patients through timely therapeutic interventions.