ACT Correlation Parameters
The correlation option determines the relationship (the “correlation”) between signal tracks. The resulting matrix of correlation coefficients can be viewed as, for example, a phylogeny tree showing the relationship between various transcription factors based on their signal values on a microarray. It does this using the method to correlate two signal tracks described in Genome Res 17: 787-97; however, it allows for simultaneous analysis of several signal tracks at once and leaves several parameters, including number of boostrap cycles and window size, flexible.
To perform the correlation calculations, the program converts the uploaded signal files into binary “hit” files where each nucleotide position is assigned to either 1 (indicating that the position is an active hit) or 0. It then calculates Pearson’s correlation coefficient for each combination of signal track uploaded; however, so that the spatial distribution of the hits is not disregarded, it does the calculation based on a sliding window in which a certain number of nucleotides at a time are aggregated and used as values for the Pearson correlation coefficient calculation.
Bootstrap cycle. Number of times to resample and recalculate the correlation coefficient when generating phylogenetic trees. This is done using phylip.
Overlap. The number of base pairs of overlap when the aggregation window “slides.” For example, in Genome Res 17: 787-97 a sliding window size of 3kb was used, with an overlap of 1.5kb.
Window size. Specifies the size of the sliding window in base pairs. For example, in Genome Res 17: 787-97 a sliding window size of 3kb was used, with an overlap of 1.5kb.
Size of heatmap picture. Length of the generated heatmap figure in inches.
The output comes in the form of several files, perhaps the most important of which is the file tracks_regions-data_matrix_original.txt which contains a matrix with correlation coefficients describing how each factor is related to every other. The rows and columns are in the order that the original files were inputted. In addition, the program generates a graphical heatmap output file (in the form of a png and pdf using R's heatmap function.) and a phylogenetic diagram showing the relationships between the inputted signal files (tree.png). The results from each individual boostrap run can be found in the bootstrap folder.