Skip to main content

Implementation of Dynamic Statistical Comparisons

Project description

PyPI version Codacy Badge

The project wiki is the main source of documentation for both developers and users of the DSC project. If you are new to the concept of DSC, it may worth reading this blog post to understand the motivation behind this project.

This work is supported by the the Gordon and Betty Moore Foundation via an Investigator Award to Matthew Stephens, Grant GBMF4559, as part of the Data-Driven Discovery program. If you have any questions or want to share some information with the developer / user community, please open a github issue.

Developer notes

Upgrading DSC to latest development version

For most users, we recommend installing the most recent stable release. If you would like to upgrade your existing installation of DSC to the most recent (unstable) development version, follow these steps.

DSC is closely developed in parallel with SoS. Therefore, the development version of DSC (maintained in the master branch of the GitHub repository) typically requires the development version of SoS.

Install the latest SoS

git clone https://github.com/vatlab/SoS.git
cd SoS
pip install -U --upgrade-strategy only-if-needed .

Install DSC from source

git clone https://github.com/stephenslab/dsc.git
cd dsc
./setup.sos

Or, if you have downloaded it before,

cd dsc
git pull
./setup.sos

Install dscrutils from source

Assuming the working directory in your R environment is the dsc repository, run the following code in R to install the latest development version of the dscrutils R package:

getwd() # Should be ... /dsc
install.packages("dscrutils",repos = NULL,type = "source")

Project maintenance

Although relatively stable and usable in practice, DSC is still actively being developed. Occasionally upgrades to the most recent version will lead to changes of file signatures that triggers rerun of existing benchmark even if they have not been changed. When this happens we will indicate in bold in our release note below that “a file signature clean up is recommended” (see release note 0.2.7.7 for example). That means after such DSC upgrades you should rerun your benchmark with --touch option to update file signatures. If possible, it is recommended that you rerun your benchmark from scratch (if resources can afford) with -s none instead of --touch to skip all existing files. We apologize for the inconveniences it incurs.

Change Log

Upcoming release

Goal for 0.2.8

  • Support for multiple output per modules.

  • Improve scripts command options.

0.2.x

0.2.7.8

  • Implement a preliminary %include feature to provide alternative code organization style.

  • Allow for ! operator in List() and Dict().

  • SoS bumped to version 0.9.13.7 for improved remote job support.

  • [minor] Various bug fixes.

0.2.7.7

  • Improvements for module with shell executables and command options.

  • Improvements for remote execution #131.

  • Improved logging.

  • Bug fixes #126, #127.

  • SoS bumped to version 0.9.13.4 for #128 and related.

A file signature clean up is recommended after this upgrade.

0.2.7.6

  • Add new feature dscrutils::shiny_plot to display simple benchmark results.

  • [minor] Display unused modules with -h option.

0.2.7.5

  • Add R / Python packages and version display with -h option.

  • Add .gitignore for cache folder when a git environment is detected.

  • SoS bumped to 0.9.13.3 that now bundles the pbs module.

0.2.7.4

  • Improved R’s sessionInfo format.

  • Bug fixes #119, #121, #122

  • [minor] Error message improvements.

0.2.7.3

  • More stringent R library and command executable check.

  • [minor] Fix a regression bug on path due to 0.2.7.2.

0.2.7.2

  • Improved Windows path support.

  • [minor] Fix a bug with nested tuple with raw().

0.2.7.1

  • Dump individual data object with scripts using dsc-query *.pkl and dsc-query *.rds.

  • [minor] Improve behavior for length 1 vector in R’s list with R() operator.

  • [minor] Various bug fixes.

0.2.7

  • #92 paired parameter input convention.

  • #90 and #93 use Rmd files as module executables.

  • #94 and #95 added DSC::replicate and command option --replicate.

  • Enhance R() operator due to use of dscrutils package. This packages is now required to parse DSC file when R modules are involved.

  • Add, by default, a variable DSC_DEBUG to output files that saves various runtime info.

  • SoS bumped to 0.9.13.2

    • Support R github package force install when version mismatches.

    • Force use pip to install local development version.

    • #97 Improved error logging and reporting behavior.

  • [minor] Revert from ruamel.yaml to yaml for better performance.

  • [minor] #96

  • [minor] #98

  • [minor] Various bug fixes.

0.2.6.5

  • Bring back partial mixed languages support. Piplines with mixed R and Python code can communicate data of limited types (recursively support array, matrix, dataframe), via ``rpy2`` as in versions prior to 0.2.5.x. Support for additional languages will be implemented on need basis with HDF5 format #86.

0.2.6.4

  • Add a dsc-io command to convert between python pickle and R RDS files – an internal command for data conversion and a test for rpy2 configuration.

0.2.6.3

  • Inline module executable via language interpreters (eg. R(), Python()).

0.2.6.2

  • [minor] Ignore leading . in file(): file(.txt) and file(txt) are equivalent.

  • [minor] Disallow derivation of modules from ensemble.

  • [minor] Various bug fixes.

0.2.6.1

  • Internally replace RDS format with HDF5 format for Python routines. Pipeline with mixed languages is now officially broken at this point until the next major release that supports ``HDF5`` in R.

  • SoS required version bumped to 0.9.12.7 for relevant upstream bug fixes for remote host computing.

  • [minor] Various bug fixes.

0.2.6

  • Bring back --host option; add a companion option --to-host to facilicate sending resources to remote computer.

  • Add --truncate switch.

  • SoS required version bumped to 0.9.12.3 for relevant upstream bug fixes.

  • [minor] Improved command interface.

0.2.5.2

  • SoS required version bumped to 0.9.12.2 for relevant upstream bug fixes.

0.2.5.1

  • Change in seed behavior: since this release seed will no longer be a DSC keyword. Users are responsible to set seeds on their own.

  • [minor] Allow for both lower case and capitalized operator names File/file, List/list, Dict/dict.

0.2.5

  • New syntax release, compatible with SoS 0.9.12.1.

  • Removed --host option due to upstream changes.

0.1.x

0.1.0

  • First release, compatible with SoS 0.6.4.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

dsc-0.2.7.8.tar.gz (132.3 kB view details)

Uploaded Source

Built Distribution

dsc-0.2.7.8-py3-none-any.whl (142.9 kB view details)

Uploaded Python 3

File details

Details for the file dsc-0.2.7.8.tar.gz.

File metadata

  • Download URL: dsc-0.2.7.8.tar.gz
  • Upload date:
  • Size: 132.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No

File hashes

Hashes for dsc-0.2.7.8.tar.gz
Algorithm Hash digest
SHA256 6bd56241993b90fcf56727636ef541048b3a16c1781fb4f8adb8a3f185c80ed5
MD5 826f7a619412805940901a2e9767515a
BLAKE2b-256 e7d64061c70635f608a4afa0424c6bdc49e8fec599f36aff5e7ede502c8f76c1

See more details on using hashes here.

File details

Details for the file dsc-0.2.7.8-py3-none-any.whl.

File metadata

File hashes

Hashes for dsc-0.2.7.8-py3-none-any.whl
Algorithm Hash digest
SHA256 917bf25fca30b687bc47169d05d251c0c354d56a066fd1588476db5c19ae732c
MD5 6c2559f96a842d82d8086b457b5ef78c
BLAKE2b-256 f0081bc8975c32af2ad5219857dd43bec0cc6e204dccb8fbc29d390fcec8dfb5

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page