Does Applicability Domain Exist in Microarray-Based Genomic Research?

Li Shao, Leihong Wu, Hong Fang et al. · 2010

Read Does Applicability Domain Exist in Microarray-Based Genomic Research? by Li Shao, Leihong Wu, Hong Fang et al. (2010) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0011055.txt.

Description

Li Shao, Leihong Wu, Hong Fang and co-authors published this genomics methods study in PLOS ONE in 2010. Models that predict patient outcomes from tumor gene profiles often fail when tested on new samples. One suspected culprit is extrapolation. That means applying a model to patients unlike those it was trained on. In drug safety science that boundary is called the applicability domain, and the authors asked whether it exists for microarray models.

They built predictive models and graded test samples by how far they sat from the training population's genomic space. Performance degraded with extrapolation degree, supporting the idea that a model's trustworthiness has borders. Shao and colleagues argue that genomic predictions should ship with a domain statement, just as chemical models do. Ignoring those borders helps explain the field's trail of failed clinical validations.

Catalog record

CREATORLi Shao, Leihong Wu, Hong Fang et al.
YEAR2010
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
LICENSECreative Commons Attribution License
RIGHTSCC BY
SOURCEPLOS ONE

Source: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0011055

Browse the full library · About the Tower of Babel