Comparison and Optimisation of Methods for Measuring Fetal DNA Fraction in Maternal Plasma Based on Next-generation Sequencing of Cell-free DNA

Objective. There were two tasks: to perform a comparative analysis of existing approaches, determining cell-free fetal DNA (cffDNA) fraction in maternal plasma; to improve and combine the considered methods using the whole genome sequencing data (one of the most important part – improvement the quality of results in low cffDNA fraction, about 4-5%). Subjects and Methods. Both basic methods for detecting a high risk group of aneuloidies and the methods (created during the development and evolution of the noninvasive prenatal tests (NIPT)) for determinations of fetal DNA fraction in blood plasma according to the next-generation sequencing data were reviewed. The study was carried out on the data of single-end whole-genomic shallow-depth sequencing (Ion Torrent), provided by the Laboratory of Molecular and Genetic Methods at Kulakov Research Center for Obstetrics, Genecology and Perinatology (Russia). Statistical analysis of the data and the search for regularities, and determination of the key features as the predictors for the machine learning models were carried out. Basic pre-existing approaches: the method based on the distribution of the lengths of free circulating fragments; seqFF; the method, using the generated full-genome nucleosome profiles. The results were compared with the cell-free fetal DNA fraction, based on the calculating of the proportion of fragments mapped to the Y chromosome, for training and qualitative evaluation of the effectiveness of the research. The ranking of the models by quality was based on the following metrics: Pearson correlation coefficient (r), mean absolute error (MAE) the mean squared error (MSE), the area under the error curve (AUC). Results. The constructed models of machine learning solve the regression problem of numerical determination of the fetal cell-free DNA fraction, as well as the classification task for determining low and high fetal fraction.