Töökindla ja teisaldatava töövoo väljatöötamine molekulaarsete tunnustega seotud geneetiliste variantide tuvastamiseks mitmetest andmestikest
Failid
Kuupäev
2019
Autorid
Ajakirja pealkiri
Ajakirja ISSN
Köite pealkiri
Kirjastaja
Abstrakt
Kvantitatiivse tunnuse lookusteks (quantitative trait locus, QTL) nimetatakse geneetilisi variante, millel on statistiline seos mõne molekulaarse tunnusega. QTL analüüs võimaldab paremini aru saada komplekshaiguseid ja tunnuseid mõjutavatest molekulaarsetest mehhanismidest. Tüüpiline QTL analüüs koosneb suurest hulgast sammudest, mille kõigi jaoks on olemas palju erinevaid tööriistu, kuid mida ei ole siiani kokku pandud ühte lihtsasti kasutatavasse, teisaldatavasse ning korratavasse töövoogu. Käesolevas töös loodud töövoog koosneb kolmest moodulist: huvipakkuva tunnuse kvantifitseerimine (i), andmete normaliseerimine ja kvaliteedikontroll (ii) ning QTL analüüs (iii). Kvantifitseerimise ja QTL analüüsi moodulite jaoks kasutasime Nextflow töövoo juhtimise süsteemi ning järgisime kõiki nf-core raamistiku parimaid praktikaid. Mõlemad töövoo moodulid on avatud lähekoodiga ning kasutavad tarkvarakonteinereid, mis võimaldab kasutajatel neid lihtsalt laiendada ning jooksutada erinevates arvutuskeskkondades. Kvaliteedikontrolli teostamiseks ning andmete normaliseerimiseks arendasime välja skripti, mis automaatselt arvutab välja erinevad kvaliteedimõõdikud ning esitab need kasutajale. Juhtprojekti raames viisime läbi geeniekspressiooni QTL analüüsi 15 andmestikus ja 40 erinevas bioloogilises kontekstis ning tuvastasime vähemalt ühe statistiliselt olulise QTLi enam kui 9000 geenile. Loodud töövoogude laialdasem kasutuselevõtt võimaldab muuta QTL analüüsi korratavamaks, teisaldatavamaks ning lihtsamini kasutatavaks.
Quantitative trait locus (QTL) analysis links variations in molecular phenotype expression levels to genotype variation. This analysis has become a standard practice to better understand molecular mechanisms underlying complex traits and diseases. Typical QTL analysis consists of multiple steps. Although a diverse set of tools is available to perform these individual analysis, the tools have so far not been integrated into a reproducible and scalable workflow that is easy to use across a wide range computational environments. Our analysis workflow consists of three modules. The analysis starts with quantification of the phenotype of interest, proceeds with normalisation and quality control and finishes with the QTL analysis. For phenotype quantification and QTL mapping modules we developed pipelines following best practices of the nf-core framework. The pipelines are containerized, open-source, extensible and eligible to be parallelly executed in a variety computational environments. For quality control module we developed a script which automatically computes the measures of quality and provides user with information. As a proof of concept, we uniformly processed more than 40 context specific groups from more than 15 studies and discovered at least one significant eQTL for more than 9000 genes. We believe that adopting our pipelines will increase reproducibility, portability and robustness of QTL analysis in comparison to existing approaches.
Quantitative trait locus (QTL) analysis links variations in molecular phenotype expression levels to genotype variation. This analysis has become a standard practice to better understand molecular mechanisms underlying complex traits and diseases. Typical QTL analysis consists of multiple steps. Although a diverse set of tools is available to perform these individual analysis, the tools have so far not been integrated into a reproducible and scalable workflow that is easy to use across a wide range computational environments. Our analysis workflow consists of three modules. The analysis starts with quantification of the phenotype of interest, proceeds with normalisation and quality control and finishes with the QTL analysis. For phenotype quantification and QTL mapping modules we developed pipelines following best practices of the nf-core framework. The pipelines are containerized, open-source, extensible and eligible to be parallelly executed in a variety computational environments. For quality control module we developed a script which automatically computes the measures of quality and provides user with information. As a proof of concept, we uniformly processed more than 40 context specific groups from more than 15 studies and discovered at least one significant eQTL for more than 9000 genes. We believe that adopting our pipelines will increase reproducibility, portability and robustness of QTL analysis in comparison to existing approaches.