Gdsctools

* NEWS:

- Elastic notebook and module implemented
- GenomicFeatures has now a compression method

* CHANGES:

- anova module was split into modules + anova so that elastic_net
module can inherit from module
- all share/data moved to gdsctools data
- add scikit-learn dependencies

* BUGS:

- Fix onevent picking in the volcano plot and use 4 digit for the FDR plot

* BUG:

- Fixes issue 127 (If MSI factor missing, the anova still tries to use it)
- Fixes issue 126 (--out-directory ignored in gdsctools-anova pipeline)
- Fixes issue 125 and 124 (HTML report links broken)

* BUG:

- Fix set_cancer_type to accept lists of tissues again

* CHANGES:

- Fixes 119 by adding more tests.
- reactivate get_significant hits functions.
- rename ANOVAResults.get_significant_hits into get_html_table

Lots of changes in this version but for users the API should be very similar.

* NEWS:

- Add a new factor called MEDIA_FACTOR. If not provided, genomic
feature matrix can populated the MEDIA_FACTOR column automatically.
- add a class COSMICInfo and a related data file called
cosmic_info.csv.gz to get information about COSMIC ids. Replaces
COSMIC class, which was removed.
- add new class GDSC to perform the entire analysis splitting data across
companies found in DrugDecode and across cancer types.

* CHANGES:

- COSMIC class removed and replaced by COSMICInfo class
- Column name convention:
- FEATURE_ANOVA_pval --> ANOVA_FEATURE_pval
- MSI_ANOVA_pval --> ANOVA_MSI_pval
- TISSUE_ANOVA_pval --> ANOVA_TISSUE_pval
- FEATURE_ANOVA_FDR_% --> ANOVA_FEATURE_FDR
- new column named ANOVA_MEDIA_pval
- to be constistent, names such as FEATURE_pos have now underscores
to separate words e.g., (FEATUREpos --> FEATURE_pos, FEATUREneg
--> FEATURE_neg, deltaMEAN --> delta_MEAN).
- refactor :mod:`gdsctools.volcano` module to use new naming convention.
- SAMPLE_NAME is not required anymore in the genomic features. This is
indeed just an annotation and is now encoded in the flat file
cosmic_info.csv.gz (see above)
- :mod:`~gdsctools.anova`, anova_results modules:
- Implement new factor (MEDIA) in the regression
- Uses new naming convention for the columns as described above
- When initialising a ANOVA instance, prints the factor that will be
included.
- add new option (set_media_factor) to populate the MEDIA column
automatically
- :mod:`~gdsctools.readers` module:
- 'Sample Name' or SAMPLE_NAME are deprecated.
There are removed from the genomic_feature matrix if found.
- Uses MEDIA_FACTOR column in addition to MSI and tissue columns
- shift attribute is now read-only and set automatically
- add a function to fill media column automatically
- print function is more verbose
- volcano: uses new naming convention for the columns as described above.
- split :mod:`~gdsctools.anova` module (create
:mod:`~gdsctools.anova_report`) (issue 98).
- :mod:`~gdsctools.readers`: improved DrugDecoder and renamed into
DrugDecode (issue 102 and 101)
- add new settings and code to apply pvalue correction at drug level
rather than global level.
- add new module to find chemblId/ChemSpider from drug name.

* NEW:

- add settings as json file in the HTML report
- ANOVAResults has now a volcano() method
- add read_settings method in ANOVA
- add code in the HTML tree directory to reproduce HTML report and results

* CHANGES:

- anova_one_drug now returns an ANOVAResults object
- Restructure data package tree directory (83)
- Default header have changed:
- COSMIC ID --> COSMID_ID
- Sample Name --> SAMPLE_NAME
- MS-instability Factor Value --> MSI_FACTOR
- Tissue Factor Value --> TISSUE_FACTOR

Previous values will still be accepted but deprecation warning added.

* BUGS:

- Fixes 89 (tight layout buggy under MAC)

* CHANGES:

- add new regression method: Ridge/Lasso/ElasticNet in
:class:`gdsctools.anova.ANOVA`
- Rename some of the settings to have a more uniform naming convention in
:class:`gdsctools.settings.ANOVASettings`
- Add new module related to fitting ot logistic function parameters
(:mod:`gdsctools.logistics`)