2016-04-18
tl;dr (summary of changes)
run function now accepts paths.# 'cherries' version
cd <path to pipelines>
flowr x=mypipeline
# 'dates' version
flowr x=<path to pipelines>/mypipeline~/.flowr.conf, one may specify their own structure - enabling flexibility.# 'cherries' version: a fixed directory structure was recommended:
~/flowr
├── conf
│ ├── flowr.conf
├── pipelines
│ ├── sleep_pipe.R
├── runs
# 'dates' version: one may change the default paths to run, config files etc using
~/.flowr.conf
# this file controls the location of these folders:
flow_base_path ~/flowr # flowr home
flow_conf_path {{flow_base_path}}/conf # path to configuration files, not required if using ~/.flowr.conf
flow_run_path ~/flowr/runs # default home of all executed flows, you may change this
flow_pipe_paths ~/flowr/pipelines,<add new paths...> # multiple paths can be specified using ","2015-12-03
tl;dr (summary of changes)
flowr.conf (modules_cmds and local_cores), after upgrading the package. Everything else is compatible with previous versions.additions/changes to flowr.conf file
New: flow_pipe_paths now supports multiple paths, seperated by comma. The fetch_pipes() would split the vector at commas.
IMP: New version needs additional components in the flowr.conf file
New: Now you can add a module_cmds variable to the config file, and this will be prefixed in all script of the pipeline. An example could be:
# version >= 0.9.8.9004
# max number of cores to use when running on a local server
local_cores 4
# default module of a pipeline
# version >= 0.9.8.9015
module_cmds ''
# examples: one may define all modules used in a pipeline here,
# further one may specify any other command which should be run before
# script executes.
#module_cmds 'module load samtools;export PATH=$PATH:/apps/bin'
addition/changes to status()
get_wds/status, so that if current wd contains a flow_details file, status is shown for this folder and not sub-folder(s).## now this works well !
flowr status x=.addition/changes to run() and rerun() functions
conf], parameter. See help(flowr::run) for more details. This enables one, to specify custom parameters used for that pipeline.rerun() supports multiple folders. Basically, one may specify a parent folder which has multiple flowr runs and ask it to re-run ALL of them again, from a specific intermediate step.New: Flowr creates a new folder if there are multiple samples in the flowmat; basically containerizes the run, keeping the logs clean and debugging life easier.
New: run() now supports, re-running as well. i.e. One would generate a new set of commands etc. but execute in the previous folder; possibly from a inter-mediate step (experimental feature).
other changes
to_flowdef() can now guess submission and dependency types (experimental feature).to_flowdef now adds a parameter nodes, to enable specifying number of nodes required per-job.opts_flow$get replaces get_opts, for reliability etc. Also this closely follows how knitr options are set.2015-10-02
status column. Specifically, this uses information from other columns and summarizes whether a specific step is pending, processing, completed or errored.================================================================================
| | total| started| completed| exit_status|status |
|:---------------|-----:|-------:|---------:|-----------:|:----------|
|001.alnCmd1 | 109| 109| 109| 0|completed |
|007.markCmd | 3| 3| 0| 0|processing |Introducing verbose levels:
One can set the level of verboseness using opts_flow$set(verbose=2). Where the level may be 0, 1, 2…. Level 1 is good for most purposes, where as, level 0 is almost silent, producing messages only when neccessary. While level 2 is good when developing a new pipeline, additional details useful for debugging are provided by level 3.
Detailed checking of flowdef
checking if required columns are present...
checking if resources columns are present...
checking if dependency column has valid names...
checking if submission column has valid names...
checking for missing rows in def...
checking for extra rows in def...
checking submission and dependency types...
jobname prev.sub_type --> dep_type --> sub_type: relationship
1: aln1_a none --> none --> scatter
2: aln2_a scatter --> none --> scatter
3: sampe_a scatter --> serial --> scatter rel: complex one:one
4: fixrg_a scatter --> serial --> scatter rel: complex one:one
5: merge_a scatter --> gather --> serial rel: many:one
6: markdup_a serial --> serial --> serial rel: simple one:one
7: target_a serial --> serial --> serial rel: simple one:one
8: realign_a serial --> burst --> scatter rel: one:many
9: baserecalib_a scatter --> serial --> scatter rel: complex one:one
10: printreads_a scatter --> serial --> scatter rel: complex one:oneSeveral detailed changes to the documentation.
2015-08-22 (0.9.7.10)
fetch_pipe()mypipeline(), mypipeline is the name of the pipeline.load_conf()fetch_pipe, and load it using as.flowdef()to_flow()kill(): now a S3 functions, and operates on both a flow object and flow_wd foldercheck(): Now works on flowdef and flowmatfetch() along with fetch_pipes() and fetch_conf() simplify finding filesparams pkg
read_sheet(), write_sheet()get_opts(), set_opts().load_conf() load_conf()2015-04-07
2015-03-16
TODO
2014-12-05
Versioning:
Loosely, following fruits A-Z.