Identification of genes with ascending or descending monotonic expression patterns over time or stages of stem cells is an important issue in time-series microarray data analysis. We propose a method named Monotonic Feature Selector (MFSelector) based on a concept of total discriminating error (DEtotal) to identify monotonic genes. It is a three-part package built in R environment (Fig. 1).
MFSelector includes a novel index, called DEtotal (total discriminating error), that can be used to measure the monotonicity of genes (i.e., with monotonically ascending/descending expression profiles) in order to identify genes/biomarkers with stronger monotonic features which may bear a correlation to stem cell development. It considers various time stages in stage order (i.e., Stage One vs. other stages, Stages One and Two vs. remaining stages and so on) and computes DEtotal of each gene. DEtotal does not make any distributional assumption about the data. MFSelector also provides the related statistical information (p- and q- value) for each monotonic gene. Moreover, in order to further identify the most monotonic gene with the same DEtotal, MFSelector also offers an additional novel index, called SVDE (sample variance for discriminating error). MFSelector can successfully identify genes with monotonic characteristics. We have demonstrated the effectiveness of MFSelector on two stem cell differentiation datasets: embryonic stem cell neurogenesis (ESCN) (Figs. 2 and 3) and embryonic stem cell vasculogenesis (ESCV) data sets (Fig. 4). Some of the monotonic marker genes such as Oct4, Nanog, Blbp, discovered from the ESCN data set exhibit consistent behavior with that reported in other studies. The detailed processes and results are explained in the publication: Discovering monotonic stemness marker genes from time-series stem cell microarray data
The role of monotonic genes found by MFSelector in either stemness or differentiation is validated using information obtained from Gene Ontology analysis and other literature. We justify and demonstrate that descending genes are involved in the proliferation or self-renewal activity of stem cells, while ascending genes are involved in differentiation of stem cells into variant cell lineages. The MFSelector R script can be downloaded from: http://microarray.ym.edu.tw/tools/MFSelector/.
|
 |
| Figure 1. The workflow of Monotonic Feature Selector (MFSelector). |
| |
 |
| Figure 2. Scatter plots of 209598_at (PNMA2) and 220668_s_at (DNMT3B) of the ESCN data set with ascending and descending profiles, respectively. (A) This is one of the top three ascending genes with DEtotal=0. (B) This is one of the top twelve descending genes with DEtotal=0. |
|
 |
| Figure 3. Scatter plots of the four monotonic genes of the ESCN data set, whose DEtotal values all are equal to one, illustrate the sample variance for discriminating error (SVDE) by adding three percent white noise to each sample for 100 simulations. (A) 227498_at (SOX6) with SVDE=0.37 (1st); (B) 221236_s_at (STMN4) with SVDE=0.91 (2nd); (C) 212614_at (ARID5B) with SVDE=2.03 (3rd); (D) 224901_at (SCD5) with SVDE=2.95 (4th). |
|
 |
| Figure 4. Scatter plots of 1552610_a_at (JAK1) and 1007_s_at (DDR1) of the ESCV data set with ascending profile and descending profile respectively. (A) This is one of the top 216 monotonically ascending genes with DEtotal=0. (B) This is one of the top 563 monotonically descending genes with DEtotal=0. |
| |
|
