Department of Computer Science: MSc Thesis Presentations

Demonstrating in silico validation analyses of qPCR assays using novel in-house software

Author: Sanni Haahti
Supervisor: Professor Juho Rousu

Abstract: The human body is constantly challenged by various pathogenic micro-organisms that exhibit diverse strategies of genetic variation to survive and procreate. This variation as well as the disproportion in overall living conditions result in continuous fluctuation in the clinical significance of the emerging variants. As a consequence, diagnostics that rely on the amplification and detection of specific gene targets, such as quantitative PCR (qPCR), require regular re-validations to sustain competence in specificity. The developing in silico methods provide a feasible alternative to excessive laboratory screening of new variants in vitro. This thesis was produced for Mobidiag Ltd, a Hologic Inc. company, as an independent project aiming to facilitate the maintenance of the assay designs updated. The work presents in-silico procedures to quantitatively validate the analytical specificities of qPCR assays with two novel in-house software, and compares the results for three multiplex qPCR test panels. The software are at diverse stages of development and thus the performances and usabilities are further qualitatively evaluated and contrasted. The results both substantiate the necessity of assay re-validations and reflect the target-dependency of the quality and amount of sequencing data; 22 % of the evaluated 36 assays were less than 98 % inclusive, while only three were found a concerning variant. The challenge of managing large volumes of data was also discerned as the existing SARS-CoV2 data was too vast to be completely retrieved and analyzed. Furthermore, the arduousness of artificially rendering biological reaction systems hinders the development of fully automated programs that execute reliable analyses without human monitoring.