| | 109 | === Comparing binding across different samples === |
| | 110 | |
| | 111 | * Method 1: Intersect bound regions (peaks) |
| | 112 | * This binary comparison asks simply, "Is each peak in sample A present in sample B (and vice versa)?" |
| | 113 | * This ignores the enrichment/score of each peak and gives each non-peak a score of 0. |
| | 114 | * Method 2: Compare affinity (number of mapped reads) across bound regions |
| | 115 | * Filter out redundant reads from each BAM file with a command like 'samtools rmdup'. |
| | 116 | * Merge peaks of all samples (and make non-redundant): |
| | 117 | * mergeBed -i Peaks_all_redundant.bed > Peaks.bed |
| | 118 | * Count reads in each sample that overlaps each peak |
| | 119 | * intersectBed -bed -wb -abam Reads_1.noDups.bam -b Peaks.bed | cut -f13-15 | mergeBed -i stdin -n > Peak_counts.reads_1.bed |
| | 120 | * Modify counts per peak to account for differing numbers of mapped reads. |
| | 121 | * Method 3: Use [[http://code.google.com/p/ngsplot/ | ngsplot]] to make stacked heatmaps of peaks for each sample |
| | 122 | |
