Simulation Modelling for Machine Learning - Identification of Single Nucleotide Polymorphisms in Human Genomes

Aug 5, 2026

Upload by Eliana

A simulation modelling approach for single nucleotide polymorphisms (SNPs) in DNA sequences is presented, generating random events using beta or normal distributions whose parameters are estimated from available experimental measurements. The method targets more accurate SNP determination under noisy sequencing conditions. Model verification and comparative analysis are conducted on GIAB consortium reference data, showing the strongest performance with Conditional Inference Trees, reaching 82.8% F1 score, outperforming binomial, entropy-based, and Fisher’s exact tests.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages5

Reading Time13 min

AccessFree to download

Introduction

Methodology

Simulation Modelling of SNPs in DNA sequences

Simulation Modelling for Machine Learning Identification of Single Nucleotide Polymorphisms in Human Genomes Mikalai M. Yatskou Deprt. of Systems Analysis and Computer Modelling Belarusian State University Minsk, 220030, Belarus [yatskou@bsu.by](yatskou@bsu.by) Elizabeth V. Smolyakova Deprt. of Systems Analysis and Computer Modelling Belarusian State University Minsk, 220030, Belarus [smolyakova580@gmail.com](smolyakova580@gmail.com) Vasily V. Grinev Deprt. of Genetics Belarusian State University Minsk, 220030, Belarus [grinev_vv@bsu.by](grinev_vv@bsu.by) Victor V. Skakun Deprt. of Systems Analysis and Computer Modelling Belarusian State University Minsk, 220030, Belarus [skakun.victor@gmail.com](skakun.victor@gmail.com) Abstract—An approach for simulation modelling of Single Nucleotide Polymorphisms (SNPs) in DNA sequences is proposed, which implements the generation of random events according to the beta or normal distributions, the parameters of which are estimated from the available experimental data. This approach improves the accuracy of determining SNPs in DNA molecules. The verification of the developed model and analysis methods was carried out on a set of reference data provided by the GIAB consortium. The best results were obtained for the machine learning model of Conditional Inference Trees – the accuracy of the SNP identification by the score F1 is 82,8 %, which is higher than those obtained by traditional SNP identification methods, such as binomial distribution, entropybased and Fisher's exact tests. Keywords — single nucleotide polymorphism, SNP identification, simulation modelling, machine learning I. INTRODUCTION Genetic polymorphism affects the human phenotype and other living organisms [1] . Single nucleotide polymorphisms (SNPs) are one of the most common types of genetic variation in the human genome. Knowledge of the genes involved in cancer development, combined with the ability of gene sequencing and bioinformatics analysis, is an important tool for screening patients at risk and assisting in genetic counseling [2] . Statistical methods of binomial distribution, entropybased, Fisher's exact tests and machine learning are applied for identifying the SNPs in humans and plants [1, 3, 4] . These methods are quite universal and simple for program implementation, however, are computationally expensive and difficult to be used in the analysis of experimental data with a high noise level and various experimental distortions, which are sources of gaps, repetitions, and other anomalous values often observed in genomic sequencing by the PacBio and Oxford Nanopore technologies [5] . In practical experimental studies, simulation modelling is used to select the most optimal SNP identification algorithm, test competing plans/methods of analysis, and evaluate the performance of specific experimental design for studying biophysical systems [6, 7] . Simulations are critical for testing methods and studying the effects of different phenotypic and genetic architectures of biological traits. Modeled genotypes and phenotypes reflect the intended understanding of the true structure of the phenotype, but do not guarantee the biological correctness of real phenotypes [8] . Simulation modelling is also used to generate training data for machine learning methods to directly identify SNP sites of various organisms from a single sequencing experiment [4] . In this case, the formation of simulated training data can have advantages in terms of accuracy and efficiency in the analysis of experimental data both with a low number of coverages and with gaps due to experimental distortions. Various approaches to mathematical modelling of genetic polymorphisms, based on accounting the parameters of experimental equipment, the use of probabilistic models and statistical approaches, and auxiliary biological information, are published elsewhere [9, 10]. However, due to complexities in the types of genetic data, modell

FAQs

How does the proposed simulation model generate data for SNP identification?

It generates random events according to beta or normal distributions, with distribution parameters estimated from experimental data.

How was the developed method verified?

Verification and analysis were performed on reference data from the GIAB consortium using comparative evaluation of existing algorithms.

Which machine learning model performed best and what accuracy was achieved?

Conditional Inference Trees achieved the best results, with SNP identification F1 score reported as 82.8%.

Aug 5, 2026

Upload by Eliana