List of available statistics and plots generated by RBPBench (rbpbench search --report):
Table: RBP motif enrichment statistics. Given a score for each genomic region (# input regions = 1), RBPbench checks whether motifs are enriched in higher-scoring regions (using Wilcoxon rank-sum test). A low Wilcoxon rank-sum test p-value for a given RBP thus indicates that higher-scoring regions are more likely to contain motif hits of the respective RBP. NOTE that if scores associated to input genomic regions are all the same, p-values become meaningless (i.e., they result in p-values of 1.0).
RBP ID | # hit regions | % hit regions | # motif hits | p-value |
---|---|---|---|---|
PUM1 | 1 | 100.00 | 1 | 1.0 |
PUM2 | 1 | 100.00 | 4 | 1.0 |
Column IDs have the following meanings: RBP ID -> RBP ID from database or user-defined (typically RBP name), # hit regions -> number of input genomic regions with motif hits (after filtering and optional extension), % hit regions -> percentage of hit regions over all regions (i.e. how many input regions contain >= 1 RBP binding motif), # motif hits -> number of unique motif hits in input regions (removed double counts), p-value -> Wilcoxon rank-sum test p-value.
RBP co-occurrences heat map.
Figure: Heat map of co-occurrences (Fisher's exact test p-values) between RBPs. Legend color: negative logarithm (base 10) of Fisher's exact test p-value. Hover box: 1) RBP1. 2) RBP2. 3) p-value: Fisher's exact test p-value (calculated based on contingency table between RBP1 and RBP2). 4) RBPs compaired. 5) Counts[]: Contingency table of co-occurrence counts (i.e., number of genomic regions with/without shared motif hits) between compaired RBPs, with format [[A, B], [C, D]], where A: RBP1 AND RBP2, B: NOT RBP1 AND RBP2 C: RBP1 AND NOT RBP2 D: NOT RBP1 AND NOT RBP2.
RBP correlations heat map.
Figure: Heat map of correlations (Pearson correlation coefficients) between RBPs. Genomic regions are labelled 1 or 0 (RBP motif present or not), resulting in a vector of 1s and 0s for each RBP. Correlations are then calculated by comparing vectors for every pair of RBPs. Legend color: Pearson correlation coefficient. Hover box: 1) RBP1. 2) RBP2. 3) RBPs compaired. 5) Counts[]: Contingency table of co-occurrence counts (i.e., number of genomic regions with/without shared motif hits) between compaired RBPs, with format [[A, B], [C, D]], where A: RBP1 AND RBP2, B: NOT RBP1 AND RBP2 C: RBP1 AND NOT RBP2 D: NOT RBP1 AND NOT RBP2.