Homologous recombination defect judgment method based on DNA sequencing data

Abstract

The invention discloses a homologous recombination defect judgment method based on DNA sequencing data. The method comprises the following steps: acquiring characteristic attributes; extracting validdata; based on a triple learning method framework, selecting three different base classifiers H1, H2 and H3 by considering good generalization ability, high accuracy and processing efficiency for multi-dimensional feature attributes; performing iterative training on the H1, H2 and H3 to obtain an extended training set, and updating a model to complete a training process; and marking a unmarked sample set U by using the trained model, and completing judgment of the HRD state according to the marking result. According to the method, the limitation that HRD state judgment is carried out by usinglocal characteristics such as a single genome instability state or a small number of genome instability states and the like is solved, the difficulty that the number of samples with clinically known HRD states is extremely small is overcome, learning of multi-characteristic attributes under existing sample data is achieved, and the performance of the HRD judgment method can be improved.

Description

technical field [0001] The invention belongs to the technical field of data science with the application background of precision medicine, and specifically relates to a method for judging homologous recombination defects based on DNA sequencing data. Background technique [0002] Homologous recombination (English name: Homologous Recombination, English abbreviation: HR) refers to the recombination between non-sister chromatids or DNA molecules containing homologous sequences on the same chromosome or within molecules, and its significance is to repair genes. damage, such as figure 1 shown. If the gene involved in homologous recombination is mutated, it may theoretically affect the homologous recombination process of DNA, and then homologous recombination defect (English name: Homologous Recombination Deficiency, English abbreviation: HRD), also known as HRD positive. In recent years, several studies have reported that HRD can not only be used for cancer diagnosis, but also...

AI Technical Summary

Problems solved by technology

The second type of method is affected by factors such as the limitations of cancer types, the baseline of mutation characteristics, and the coverage of point mutation sets. It is still immature and is still in the stage of scientific research and exploration

[0007] Although the first method has been applied in clinical diagnosis, there are still some defects in the clinical practice in our country, the main problem is that the judgment threshold is difficult to determine and optimize

Foundation Medicine's FoundationFocus CDxBRCA LOH assay determines whether HRD is positive based on the LOH value of the tumor genome, but experiments have found that some ovarian cancer patients with low LOH are also HRD positive

MyChoice HRD assay from Myraid is based on high HRD score and BRCA mutation to judge patients as HRD positive, low HRD score and BRCA no mutation to judge patients as HRD negative, and if HRD score fails and BRCA wild type, HRD status cannot be determined

However, changes in LOH and copy number among the features of the HRD score may be caused by rearrangements of chromosomes between non-homologous recombination repairs

In summary, the two existing methods cannot perfectly solve the problem of using local features such as genome instability to determine the HRD status, and there are large errors, and the effect needs to be improved urgently

Images