Biology subjects

Sztromwasser, P.

Publications and source records attributed to Sztromwasser, P..

3 recordsLinked to original sources

bioRxiv2022

DBFE: Distribution-based feature extraction from copy number and structural variants in whole-genome data

MotivationWhole-genome sequencing has revolutionized biosciences by providing tools for constructing complete DNA sequences of individuals. With entire genomes at hand, scientists can pinpoint DNA fragments responsible for different cancers and predict patient responses to cancer treatments. However, the sheer volume of whole-genome data makes it difficult to encode the characteristics of genomic variants as features for machine learning algorithms. ResultsWe present three feature extraction methods that facilitate classifier learning from distributions of genomic variants. The proposed approaches use binning, clustering, and kernel density estimation to produce features that discriminate between two groups of patients. Experiments on genomes of 219 ovarian, 61 lung, and 929 breast cancer patients show that the proposed approaches automatically identify genomic biomarkers associated with cancer subtypes and clinical response to oncological treatment. Finally, we show that the extracted features can be used alongside unsupervised learning methods to analyze genomic samples. AvailabilityThe source code of the presented algorithms and reproducible experimental scripts are available on Github at https://github.com/MNMdiagnostics/dbfe Contactmaciej.piernik@cs.put.poznan.pl

genomics↗

bioRxiv2021

The Thousand Polish Genomes Project- a national database of Polish variant allele frequencies

Although Slavic populations account for over 3.5% of world inhabitants, no centralized, open source reference database of genetic variation of any Slavic population exists to date. Such data are crucial for either biomedical research and genetic counseling and are essential for archeological and historical studies. Polish population, homogenous and sedentary in its nature but influenced by many migrations of the past, is unique and could serve as a good genetic reference for middle European Slavic nations. The aim of the present study was to describe first results of analyses of a newly created national database of Polish genomic variant allele frequencies. Never before has any study on the whole genomes of Polish population been conducted on such a large number of individuals (1,079). A wide spectrum of genomic variation was identified and genotyped, such as small and structural variants, runs of homozygosity, mitochondrial haplogroups and Mendelian inconsistencies. The allele frequencies were calculated for 943 unrelated individuals and released publicly as The Thousand Polish Genomes database. A precise detection and characterisation of rare variants enriched in the Polish population allowed to confirm the allele frequencies for known pathogenic variants in diseases, such as Smith-Lemli-Opitz syndrome (SLOS) or Nijmegen breakage syndrome (NBS). Additionally, the analysis of OMIM AR genes led to the identification of 22 genes with significantly different cumulative allele frequencies in the Polish (POL) vs European NFE population. We hope that The Thousand Polish Genomes database will contribute to the worldwide genomic data resources for researchers and clinicians.

genomics↗

bioRxiv2021

Accuracy of somatic variant detection workflows for whole genome sequencing experiments

Whole genome sequencing (WGS) becomes increasingly important for advancing personalized cancer care, driving not only basic science studies but also entering into clinical applications. Translating raw WGS data into the right clinical decision requires high accuracy of somatic variant detection, therefore novel data analysis methods have to be carefully evaluated. In this work we tested the performance of well-established somatic variant detection workflows: GATK, CPG-WGS, DRAGEN and Strelka2. By utilizing both real data, with well-defined mutations, and synthetic mutations spiked-in into real data, we were able to assess sensitivity and precision of each workflow, for various coverage and tumor purity levels. Individual tools excelled in different evaluation approaches, however the results demonstrated that DRAGEN has the highest overall performance when sensitivity is preferred over precision, and the opposite is true for CGP-WGS. The differences in results obtained using synthetic and real datasets, indicate that benchmarks based only on a single reference set may provide an incomplete picture.

bioinformatics↗