Table 3.
Performance of iterative correction on IMR90 data
| HiC-Pro – Iced (dense – 1 CPU) | HiC-Pro – Iced (sparse – 1 CPU) | HiCorrector – MES (dense – 1 CPU) | HiCorrector – MEP (dense – 8 CPUs) | |
|---|---|---|---|---|
| IMR90 1Mbp | 00:00:12 | 00:00:25 | 00:00:25 | 00:00:06 |
| IMR90 500 kbp | 00:00:40 | 00:01:30 | 00:02:15 | 00:00:22 |
| IMR90 150 kbp | - | 00:04:28 | 00:13:21 | 00:03:10 |
| IMR90 40 kbp | - | 00:07:19 | 02:35:34 | 00:35:43 |
| IMR90 2 0kbp | - | 00:08:36 | 12:57:17 | 02:34:05 |
HiC-Pro is based on a fast implementation of the iterative correction algorithm. We therefore compare our method with the MES (Memory-Efficient Sequential) and MEP (Memory-Efficient Parallel) algorithms of the HiCorrector software [20] for Hi-C data normalization (hours:minutes:seconds). All algorithms were terminated after 20 iterations (see Additional file 1 for details)