Charting Gene Expression Equivalence in Human Lung Tissue Using the (Two One-Sided Tests) TOST Procedure
Poster Number
2A
Research or Creativity Area
Engineering & Computer Science
Abstract
Embarking on an unprecedented voyage into the molecular intricacies of human lung tissue, this study harnesses the Two One-Sided Tests (TOST) for equivalence—a rigorous statistical technique typically underexplored in the realm of RNA sequencing (RNA-seq). This research deciphers the gene expression profiles across 10,360 cells and 13 cell types from a human lung dataset, dissecting similarities where subtleties matter most.
Central to our approach is the TOST procedure, which transcends conventional binary outcomes of significance testing. By embracing the challenge to validate equivalence within a predetermined boundary, we redefine the null hypothesis in a dual configuration: we seek to confirm that the difference in mean gene expressions between cell types is neither greater than some value nor less than that value. Only by satisfying both conditions do we ascribe equivalence, ensuring robustness against type I errors without necessitating multiple testing corrections.
This delicate balance between rejecting non-equivalence and asserting similarity paints a new landscape for computational biology. With this innovative application of TOST, we elevate RNA-seq data analysis from its traditional interpretations to a tripartite classification—differential, equivalent, or inconclusive. This leads the way for a nuanced atlas of gene behavior. Our findings open a new frontier for understanding cellular dynamics with the broader understanding of complex diseases.
Location
Don and Karen DeRosa University Center (DUC) Poster Hall
Start Date
27-4-2024 10:30 AM
End Date
27-4-2024 12:30 PM
Charting Gene Expression Equivalence in Human Lung Tissue Using the (Two One-Sided Tests) TOST Procedure
Don and Karen DeRosa University Center (DUC) Poster Hall
Embarking on an unprecedented voyage into the molecular intricacies of human lung tissue, this study harnesses the Two One-Sided Tests (TOST) for equivalence—a rigorous statistical technique typically underexplored in the realm of RNA sequencing (RNA-seq). This research deciphers the gene expression profiles across 10,360 cells and 13 cell types from a human lung dataset, dissecting similarities where subtleties matter most.
Central to our approach is the TOST procedure, which transcends conventional binary outcomes of significance testing. By embracing the challenge to validate equivalence within a predetermined boundary, we redefine the null hypothesis in a dual configuration: we seek to confirm that the difference in mean gene expressions between cell types is neither greater than some value nor less than that value. Only by satisfying both conditions do we ascribe equivalence, ensuring robustness against type I errors without necessitating multiple testing corrections.
This delicate balance between rejecting non-equivalence and asserting similarity paints a new landscape for computational biology. With this innovative application of TOST, we elevate RNA-seq data analysis from its traditional interpretations to a tripartite classification—differential, equivalent, or inconclusive. This leads the way for a nuanced atlas of gene behavior. Our findings open a new frontier for understanding cellular dynamics with the broader understanding of complex diseases.
