'misha' package is intended to help users to efficiently analyze genomic data achieved from various experiments.
Details
For a complete list of help resources, use library(help = "misha").
The following options are available for the package. Use 'options' function to alter the value of the options.
| NAME | DEFAULT | DESCRIPTION |
| gmax.data.size | AUTO | Maximal number of records (rows of a result, values, bases) |
| a call holds in memory: gextract, gscreen and the like fail above it, | ||
| and gquantiles samples. A count, not a size in bytes. | ||
| Auto-configured from system RAM and cores to the number | ||
| min(RAM in bytes * 0.7 / (70% of the cores), 1e10). | ||
| Inf or unset means no limit; NA or a negative value is an error; | ||
| a value below 1 counts as 1. | ||
| gbig.intervals.size | 1000000 | Threshold for converting interval sets to disk-based |
| "big" format. When interval count exceeds this, intervals | ||
| are stored per-chromosome instead of all in memory. | ||
| Note: Independent of gmax.data.size (different purposes). | ||
| gmax.mem.usage | AUTO | Auto-configured to max(10000000, 80% of RAM in KB). |
| Maximal memory consumption of all child | ||
| processes in KB before the limiting algorithm is invoked. | ||
| 0 runs the child processes one at a time; Inf means no limit; | ||
| NA or a negative value is an error. When gmax.data.size sets no | ||
| limit, it also bounds the shared memory for the results of | ||
| child processes (half the RAM at most) and gquantiles' samples. | ||
| gmax.processes | AUTO | Auto-configured to min(32, 70% of CPU cores). |
| Maximal number of processes for multitasking. Each is a fork of | ||
| the R session; from a session holding 20 GB in 4 KB pages | ||
| a fork took 0.7 s (see the user manual). | ||
| gmax.processes2core | 2 | Maximal number of processes per CPU core for multitasking |
| gseq.extract.probe.usec | 100 | gseq.extract times its first reads |
| and distributes across processes when the measured cost per | ||
| interval reaches this many microseconds. 0 always distributes, | ||
| a large value never does. | ||
| gmin.scope4process | 10000 | Minimal scope range (for 2D: surface) assigned to a process |
| in multitasking mode. | ||
| gbuf.size | 1000 | Size of track expression values buffer. |
| gmultitask.max.records.factor | 64 | Multiplier to inflate multitask |
| max_records estimates to avoid under-allocation. | ||
| gtrack.chunk.size | 100000 | Chunk size in bytes of a 2D track. If '0' chunk size |
| is unlimited. | ||
| gtrack.num.chunks | 0 | Maximal number of 2D track chunks simultaneously stored |
| in memory. | ||
| gmultitasking | TRUE | Enable/disable automatic parallelization. Some functions |
| may choose single-threaded mode for very small workloads. | ||
| gpermissions.umask | "0007" | umask applied around misha's own writes |
| into a database: 660 files, 770 directories - group-writable, | ||
| no world access - matching what the C++ layer enforces. Set to | ||
| "0002" for the world-readable 664/775 that misha produced before | ||
| 5.11.21, or to NULL to leave the process umask untouched. The | ||
| umask is restored as soon as the write returns. |
More information about the options can be found in 'User manual' of the package.
Author
Maintainer: Aviezer Lifshitz aviezer.lifshitz@weizmann.ac.il
Authors:
Aviezer Lifshitz aviezer.lifshitz@weizmann.ac.il
Misha Hoichman misha@hoichman.com
Eitan Yaffe eitan.yaffe@weizmann.ac.il
Amos Tanay amos.tanay@weizmann.ac.il
Other contributors:
Weizmann Institute of Science [copyright holder]