DNase I Peak Callers Benchmarked in This Study.
2014-05-08T03:20:35Z (GMT) by
<p>This table shows a number of properties of the four peak calling algorithms used in this study, highlighting some of the differences (as evaluated on a Intel(R) Xeon(R) CPU @ machine). The CPU Time in seconds was obtained when running each algorithm on the K562 data (for the other two cell type comparable figures are obtained, data not shown). The Max Memory column shows the maximum memory used by the algorithm. The numbers in parentheses show the maximum number of processes used when generating this data. For MACS and ZINBA control data is optional and is believed to improve the accuracy of the algorithm, but in this study all algorithms have been run without control data sets. Hotspot and ZINBA require mappability data and both algorithms provide it for a set of specific lengths including 36 bp. The core Hotspot algorithm is implemented in , however the Hotspot pipeline involves <i>R</i>, python and bash scripts. ZINBA comes as an <i>R</i> package having its core implemented in <i>C</i>.</p>