3.4. Creation of age-representative samples: AgeCollapsed
Last updated
Last updated
All studies were transformed into a new dataset called ageCollapsed, in which all age-matched samples, regardless of study origin, were combined into a single representative sample of that age. Once the studies were aggregated through their common transcripts, the expression values were corrected for the batch effect (in this case, the studies) using the Combat package (JOHNSON; LI; RABINOVIC, 2007) from R. Then, it was combined age-matched samples are measured by the median expression value of each transcript. The median was used as it is a metric less susceptible to outliers. Once the age samples were collapsed, the age-representative samples were reordered in ascending order and each transcript treated as a time series, as exemplified in Figure 3.7. All subsequent analyzes were performed on ageCollapsed.