All sequencing tracks were made using the Washington University
Epigenome Browser. Sequencing-coverage tracks that were used to compare DNase,
standard ATAC-seq, and Omni-ATAC-seq data were generated by subsampling 60
million reads from an aligned and de-duplicated BAM file that had not been
additionally filtered. These equal-depth BAM files were then converted to bigwig
for visualization. For comparisons involving DNase-seq, all ATAC-seq reads were
trimmed to 36 bp to match the single-end 36-bp sequencing reads used in DNase
before alignment. The y-axis scale for all sequencing tracks
was set to range from 0 to the maximum height among the three data sets. In this
way, the heights of the tracks were comparable across techniques, as they were
derived from the same number of equal-length input reads. Sequencing tracks that
were used to compare 500-cell Omni-ATAC to 500-cell standard ATAC-seq data in
GM12878 cells were not normalized. In these visualizations, all pass-filter
reads were used to generate sequencing tracks under the assumption that these
libraries were sequenced to near-full depth. This assumption is necessary due to
differences in the library sizes of the Omni-ATAC and standard ATAC-seq 500-cell
libraries. Sequencing tracks related to frozen human brain tissue were all
normalized by the total number of reads in the peaks.
Epigenome Browser. Sequencing-coverage tracks that were used to compare DNase,
standard ATAC-seq, and Omni-ATAC-seq data were generated by subsampling 60
million reads from an aligned and de-duplicated BAM file that had not been
additionally filtered. These equal-depth BAM files were then converted to bigwig
for visualization. For comparisons involving DNase-seq, all ATAC-seq reads were
trimmed to 36 bp to match the single-end 36-bp sequencing reads used in DNase
before alignment. The y-axis scale for all sequencing tracks
was set to range from 0 to the maximum height among the three data sets. In this
way, the heights of the tracks were comparable across techniques, as they were
derived from the same number of equal-length input reads. Sequencing tracks that
were used to compare 500-cell Omni-ATAC to 500-cell standard ATAC-seq data in
GM12878 cells were not normalized. In these visualizations, all pass-filter
reads were used to generate sequencing tracks under the assumption that these
libraries were sequenced to near-full depth. This assumption is necessary due to
differences in the library sizes of the Omni-ATAC and standard ATAC-seq 500-cell
libraries. Sequencing tracks related to frozen human brain tissue were all
normalized by the total number of reads in the peaks.
