Skip to main content
Fig. 3 | BMC Bioinformatics

Fig. 3

From: Practical impacts of genomic data “cleaning” on biological discovery using surrogate variable analysis

Fig. 3

The biological model limits the scope of biological questions that can be asked. Defining a biological model that only preserves the effect of treatment obscures other true biological effects. The RPS4Y1 gene is differentially expressed by sex (a). However, when the biological model passed to SVA does not include sex (i.e. using the treatment only model used in Figs. 1 and 2), the effect of sex at this gene is not apparent (b). When the effects defined in SVA include sex, the difference by sex is preserved in the data (c). Similarly, with GSTT1, copy number variation has a large impact on gene expression (d) which is removed by SVA under a treatment-only biological model (e). Including a term for GSTT1 copy number in the biological model passed to SVA preserves the effect (f). Individual cell lines are represented on the X axis. Gene expression on the Y-axis is depicted in quantile normalized, log2-scale intensities

Back to article page