Table 3.
Example 1 Low complexity/Sparse read coverage | Example 2 Intermediate complexity/Low background | Example 3 High complexity/High background | Merged Replicates** | ||
---|---|---|---|---|---|
Final library concentration (μM) | 1.3 | 6.6 | 33.9 | N/A | |
Total reads | 59,241,310 | 87,868,338 | 81,734,480 | ||
% Uniquely mapped | 62.8716 | 69.0193 | 64.6791 | ||
non duplicate fraction | 0.0757579 | 0.133552 | 0.463592 | ||
CpGs (coverage ≥ 5) | 270,381 | 1,302,954 | 2,089,067 | 3,379,747 | |
Genrich | p-value (< 0.05) | q-value (< 0.05) | |||
Peaks called | 456,542 | 169,122 | 80,894 | 69,886 | |
FRiP* | 0.476 | 0.428 | 0.205 | 0.217 | |
p-value (<0.01) | q-value (< 0.01) | ||||
Peaks called | 34,718 | 60,870 | 29,990 | 50,199 | |
FRiP* | 0.201 | 0.346 | 0.137 | 0.185 | |
p-value (<0.005) | q-value (< 0.005) | ||||
Peaks called | 57,944 | 73,068 | 31,091 | 43,792 | |
FRiP* | 0.174 | 0.315 | 0.119 | 0.173 | |
MACS2 *** | q-value (<0.05) narrowPeak | N/A | |||
Peaks called | 15,322 | 64,127 | 76,310 | ||
FRiP* | 0.08 | 0.277 | 0.131 | ||
q-value (<0.05) broadPeak | |||||
Peaks called | 15,294 | 58,423 | 64,903 | ||
FRiP* | 0.085 | 0.232 | 0.138 | ||
q-value (<0.005) narrowPeak | |||||
Peaks called | 9,780 | 63,903 | 57,829 | ||
FRiP* | 0.063 | 0.243 | 0.115 | ||
q-value (<0.05) broadPeak | |||||
Peaks called | 9,615 | 55,782 | 48,217 | ||
FRiP* | 0.066 | 0.267 | 0.12 |
Fraction of reads in called peak regions. All reads mapping to non-mitochondrial chromosomes (non-chrM) in peak regions are divided by the total number of total non-chrM reads to evaluate the signal to noise ratio in ATAC-seq data. Background descriptions are based on this ratio. Generally, FRiP scores correlate with the number of regions called, and scores ≥ 0.2 are considered acceptable for standard ATAC-seq52, although ATAC-Me FRiP scores may differ. We have filtered out mitochondrial reads for this calculation as we do not call peaks on mitochondrial fragments.
Genrich can incorporate replicates into peak calling analysis and the resulting merged data is produced using the arguments described in “Data Processing”. MACS2 does not have this option. As a consequence, merged replicate data for MACS2 peaks will depend on the approach chosen for determining consensus peaks.
MACS2 has the option of calling “broad” or “narrow” peaks. We provide both here.