LTLA / HVGDetection2018

Justification for the variance modelling and HVG detection methods in scran.

Description of the HVG machinery in scran

This directory contains some files describing the HVG detection machinery in scran. In theory/:

• description.tex focuses on the theoretical basis behind trendVar and decomposeVar.
• comparison.tex considers the differences between the possible HVG detection methods in scran.

In real/:

• fitTrendVar_test.Rmd tests the behaviour of the fitTrendVar() on a range of scenarios.
• fitTrendCV2_test.Rmd tests the behaviour of the fitTrendCV2() on a range of scenarios.

In simulations/:

• power contains simulation scripts to compare the performance of decomposeVar with technicalCV2 and improvedCV2 for detecting HVGs. Each simulation script describes a different type of variability, though this is not particularly important for gene-wise testing.
• alpha assesses the type I error rate for the tests in decomposeVar, technicalCV2 and improvedCV2.
• filter examines the motivation for the default filter threshold in trendVar.