CHANGELOG.rst

FUNPACK changelog
=================


1.8.0 (Under development)
-------------------------


Added
^^^^^


* New ``take`` option to the :func:`.binariseCategorical` processing function,
  which allows the generated columns to contain values from another column,
  instead of containing binary labels.


1.7.1 (Thursday 30th January 2020)
----------------------------------


Added
^^^^^


* New built-in ``ukb`` configuration, which applies NA insertion, categorical
  recoding, and child value replacement rules from the ``fmrib`` configuration.


Fixed
^^^^^

* Fixed a bug which arose from combining the ``--import_all`` and ``--column``
  options.


1.7.0 (Friday 24th January 2020)
--------------------------------


Added
^^^^^


* New ``--index_visits`` option, which re-arranges variables with separate
  columns per visit into single columns indexed by both subject ID and visit.


Changed
^^^^^^^


* The ``--index`` option now supports specification of multiple index columns
  for each input file.
* The :func:`.fileinfo.has_header` function has been modified to be more
  lenient.
* The :mod:`.importing` module has been internally refactored to improve
  code cleanliness.
* Various minor internal API changes.
* The :func:`.removeIfRedundant` processsing function will now drop columns
  which are redundant with respect to other columns which have already been
  dropped.
* Update to the FMRIB configuration (handling variable `6150
  <https://biobank.ctsu.ox.ac.uk/crystal/field.cgi?id=6150>`_).
* The ``'codingdesc'`` metaproc function takes into account possible
  categorical recodings when retrieving the description for a particular
  value.


Fixed
^^^^^


* The :func:`.removeIfRedundant` processsing function was unnecessarily
  evaluating column pairs more than once, when run in parallel.


1.6.0 (Wednesday 11th December 2019)
------------------------------------


Added
^^^^^


* Non-numeric variables can now be used in conditional expressions, e.g.
  ``'v41202 == "A009"'`. Within such expressions, the value must be contained
  within single or double quotes.
* New ``contains`` operator, for use within conditional expressions to test
  presence of sub-strings.


Changed
^^^^^^^


* Parallelisation is now disabled by default, and must be explicitly enabled
  via the ``--num_jobs`` option. This is done in the ``fmrib`` configuration.
* Subject inclusion expressions are now evaluated during, rather than after,
  data import. They are now therefore performed in parallel on different
  chunks of the input file(s) (when parallelisaton is enabled).


1.5.0 (Monday 9th December 2019)
--------------------------------


Added
^^^^^


* New :func:`.util.wc` function to count the rows (lines) of a file;
  this is simply a wrapper around the UNIX ``wc`` tool.
* New :func:`.util.cat` function to concatenate multiple files together;
  this is simply a wrapper around the UNIX ``cat`` tool.
* New :func:`.util.inMainProcess` function so a process can determine whether
  it is the main process or a worker process.
* New :meth:`.DataTable.subtable` and :meth:`.DataTable.merge` methods, to aid
  in passing data to/from worker processes.
* Processing functions can now be specified to run independently on a subset
  of variables by using ``'independent'`` in the variable list.
* New ``any`` and ``all`` operations which can be used in conditional
  statements to control how the conditional results are combined across
  multiple columns for one variable. These can be used with the ``--subject``
  option.


Changed
^^^^^^^


* FUNPACK will now parallelise tasks by default; previously it would only
  parallelise tasks if ``--low_memory`` mode were selected.
* The data import stage is parallelised by using multiple processes to read
  different chunks of the input file(s), and then concatenating the resulting
  ``pandas.DataFrame`` objects afterwards.
* Cleaning functions are executed on each variable in parallel.
* Each processing step is executed in parallel where possible
  (e.g. ``independent`` processes), but processing steps are still executed
  sequentially.  New columns created by processing functions are saved to
  disk, and re-loaded by the main process, rather than being passed back to
  the main process via inter-process communication.
* The ``removeIfRedundant`` process now compares pairs of columns in parallel.
* The data export stage is parallelised by writing chunks of rows to different
  files, and then concatenating them into a single output file afterwards.
* The ``--variable``, ``--subject`` and ``--exclude`` options now accept
  comma-separated mixtures of IDs and MATLAB-style ranges.
* Updates to FMRIB categories.
* Updates to FMRIB processing rules, to take advantage of parallelism.
* The ,:mod:`icd10` module must now be initialised via the
  :func:`.icd10.initialise` function, when it is to be used in a multiprocessing
  context. This is not necessary when ``funpack`` is configured to not
  parallelise tasks (e.g. with ``--num_jobs 1``).


Deprecated
^^^^^^^^^^


* The ``--low_memory`` and ``--work_dir`` options have been deprecated, and no
  longer have any effect. The :mod:`.storage` module is no longer used, but is
  still present for possible future usage.


1.4.5 (Thursday 5th December 2019)
----------------------------------


Changed
^^^^^^^


* The ``funpack_demo`` notebook is now executed from a temporary directory, so
  it does not require write-permissions to the FUNPACK installation directory.


Fixed
^^^^^


* Fixed a bug where non-numeric variables (e.g. `41271
  <https://biobank.ctsu.ox.ac.uk/crystal/field.cgi?id=41271>`_ ) were being
  interpreted by ``pandas`` as being numeric.


1.4.4 (Friday 15th November 2019)
---------------------------------


Changed
^^^^^^^


* Updates to the FMRIB categories and configuration.


1.4.3 (Monday 11th November 2019)
---------------------------------


Changed
^^^^^^^


* Updated internal variable and data coding tables to the latest available from
  the UK Biobank showcase.
* Increased the file sample size used by :func:`.fileinfo.sniff`.


1.4.2 (Tuesday 6th August 2019)
-------------------------------


Changed
^^^^^^^


* Minor changes to the FMRIB configuration.


1.4.1 (Monday 8th July 2019)
----------------------------


Added
^^^^^


* New ``--trust_types`` command-line flag which tells FUNPACK to assume that
  the data in known-to-be-numeric columns is parseable (i.e. that there are no
  bad/unparseable values). This option improves import performance, but at the
  cost of causing FUNPACK to crash if the assumption is not true.


1.4.0 (Sunday 7th July 2019
---------------------------


Added
^^^^^


* Added a new ``InternalType`` column to the variable table, which can be used
  to specify the type to use internally for a given variable
  (e.g. ``float64``).  This is so that the default type of ``float32`` can be
  overridden for specific variables for which this is problematic, such as
  variable :ref:`20003
  <https://biobank.ctsu.ox.ac.uk/crystal/field.cgi?id=20003>`_. This column is
  initially populated from ``funpack/data/type.txt``.
* New :mod:`funpack.coding` module, for retrieving descriptive information
  about data codings. The information is stored in the
  ``funpack/data/coding/``directory.  Hierarchical data codings are still
  accessed via the :mod:`.hierarchy` module.
* New :func:`hierarchicalDescriptionFromCode`,
  :func:`hierarchicalDescriptionFromNumeric`, and
  :func:`codingDescriptionFromValue` metaprocessing functions.



Changed
^^^^^^^

* The hierarchical coding name no longer needs to be specified when using the
  :func:`.codeToNumeric` cleaning function - the coding is automatically looked
  up.
* Variable 4288 has been moved from ``cognitive phenotypes`` to
  ``miscellaneous`` in the FMRIB categories.
* Variable 20003 is now binarised in the FMRIB categories.
* Changed the meta-processing function signature - these functions are now
  passed the :class:`.DataTable` and variable ID, in addition the value.


Fixed
^^^^^


* Now using an internal type of ``float64`` for variable 20003, as it
  potentially has values which cannot be represented in ``float32``.


Deprecated
^^^^^^^^^^


* Deprecated the `xDescriptionFromCode` and `xDescriptionFromNumeric`
  metaprocessing functions.


1.3.2 (Tuesday 4th June 2019)
-----------------------------


Changed
^^^^^^^


* Minor adjustments to the FMRIB categories.


1.3.1 (Thursday 30th May 2019)
------------------------------


Changed
^^^^^^^


* Updates to documentation.


1.3.0 (Wednesday 29th May 2019)
-------------------------------


Added
^^^^^


* New :func:`.codeToNumeric` cleaning function, for transforming hierarhical
  variable codes.
* New :func:`.hierarchy.codeToNumeric` and
  :func:`.hierarchy.numericToCode` functions.
* New meta-process functions for generating descriptions for ICD9, OPCS3 and
  OPCS4 hierarchical variables.
* Variable, data coding, processing, category and type files in the
  ``funpack/config`` directory can be specified on the command line and in
  configuration files as relative paths, and using a "dot" syntax, e.g.
  ``fmrib/categories.tsv``, or ``fmrib.categories``.


Changed
^^^^^^^


* Built-in cleaning and processing rules are no longer applied by default -
  they are now a part of the built-in ``fmrib`` configuration, and can be
  applied via ``-cfg fmrib``.
* Updates to built-in ``fmrib`` processing.
* The ``flattenHierarchical`` processing function accepts a ``name`` argument,
  allowing the hierarchical data type name to be specified.  If not provided,
  the type is inferred from the variable ID if possible.


Fixed
^^^^^


* Fixed a bug where a processing step attempted to add a new column with
  the same name as an existing one.


Deprecated
^^^^^^^^^^


* The :func:`.convertICD10Codes` cleaning function has been replaced by
  the new :func:`.codeToNumeric` function, which can be used with any
  hierarchical variable.
* The :func:`.icd10.codeToNumeric` and :func:`.icd10.numericToCode` functions
  have been replaced by the :func:`.hierarchy.codeToNumeric` and
  :func:`.hierarchy.numericToCode` functions.
* The :func:`.loadDefaultTables` function is obsolete and has been deprecated.



1.2.1 (Tuesday 28th May 2019)
-----------------------------


Changed
^^^^^^^


* Minor changes to built-in variable categories.


1.2.0 (Saturday 25th May 2019)
------------------------------


Added
^^^^^


* New ``--summary_file`` option, which exports a summary of the
  cleaning/processing steps that have been applied to each variable.


Changed
^^^^^^^


* Built-in recoding, NA insertion, and child value replacement rules have
  been revised and updated.


1.1.4 (Friday 17th May 2019)
----------------------------


Changed
^^^^^^^


* Changed default processing rules so a column with standard deviation less
  than ``1e-6`` is deemed sparse, and dropped.



1.1.3 (Thursday 16th May 2019)
------------------------------


Changed
^^^^^^^


* The :func:`.isSparse` function has been changed so that, when the ``mincat``
  or ``maxcat`` tests are specified as proportions, they are applied relative
  to the number of *non-missing* data points, rather than the total number of
  data points.


1.1.2 (Thursday 16th May 2019)
------------------------------


Fixed
^^^^^


* Fixed a bug in :func:`.flattenHierarchical` with respect to handling missing
  values.


1.1.1 (Wednesday 15th May 2019)
-------------------------------


Fixed
^^^^^


* Changed the :func:`.isSparse` function to avoid issues with non-numaric
  data.


1.1.0 (Tuesday 14th May 2019)
-----------------------------


Changed
^^^^^^^


* The ``--visit``/``-vi`` command line option will no longer be applied to
  variables which do not have an `instancing
  <https://biobank.ctsu.ox.ac.uk/crystal/schema.cgi?id=9>`_ code 2.
  This is implemented in the :func:`.keepVisits` function.
* The :func:`.remove` and :func:`.keepVisits` function signatures have changed
  - they now require the variable table to be passed in as the first argument.


1.0.2 (Tuesday 14th May 2019)
-----------------------------


Changed
^^^^^^^


* The :func:`.removeIfSparse` processing function accepts an ``ignoreType``
  parameter which forces all tests to be run, regardless of the variable type.


Fixed
^^^^^


* The :func:`.isSparse` function was skipping the ``mincat``/``maxcat`` tests
  for non-numeric categorical variables.


1.0.1 (Friday 9th May 2019)
---------------------------


Changed
^^^^^^^


* Python package name changed from ``fmrib_unpack`` to ``fmrib-unpack``.


1.0.0 (Friday 9th May 2019)
---------------------------


Changed
^^^^^^^


* ``ukbparse`` is now called ``FUNPACK`` - the *FMRIB UKBiobank Normalisation,
  Processing And Cleaning Kit*.


Removed
^^^^^^^


* The ``ukbparse_htmlparse``, ``ukbparse_join``, and
  ``ukbparse_compare_tables`` scripts have been removed.
* The ``ukbparse.icd10.readICD10CodingFile`` function and
  ``ukbparse.icd10.ICD10Hierarchy`` class have been removed (their
  functionality was replaced by the :mod:`.hierarchy` module)
* The :func:`.processing_functions.removeIfSparse` and
  :func:`.processing_functions_core.removeIfSparse` functions no longer accept
  an ``absolute`` argument.



0.21.1 (Thursday 8th May 2019)
------------------------------


Changed
^^^^^^^


* Addd categories 1, 2 and 99 to the ``fmrib`` configuration.



0.21.0 (Thursday 8th May 2019)
------------------------------


Added
^^^^^


* :class:`.Column` objects now have a ``metadata`` attribute which may be used
  in the column description (if the ``--description_file`` option is used).
  Processing functions can set the metadata for newly added columns.
* New ``metaproc`` plugin type to manipulate column metadata.
* All processing functions accept a ``metaproc`` argument, allowing a
  ``metaproc`` function to be applied to any column metadata that is returned
  by the processing function..


Changed
^^^^^^^


* The :func:`.binariseCategorical` function sets the categorical value as
  column metadata on the new binarised columns.


0.20.1 (Wednesday 8th May 2019)
-------------------------------


Fixed
^^^^^


* Fixed some typos in the ``README`` file.


0.20.0 (Tuesday 7th May 2019)
-----------------------------


Added
^^^^^


* The :func:`.isSparse` and :func:`.removeIfSparse` functions accept
  a new option, ``mincat``, which allows a categorical to be deemed sparse
  if the size of its smallest category is below a given threshold.
* New ``--description_file`` option which, for UK BioBank data, saves the
  description for each column to a text file.


Changed
^^^^^^^


* The ``absolute`` parameter to the :func:`.isSparse` and
  :func:`.removeIfSparse` functions is deprecated. Instead, they now accept
  ``abspres`` and ``abscat`` arguments, allowing the
  absoluteness/proportionality of the ``minpres`` and ``mincat``/``maxcat``
  options to be specified separately.
* Changed default processing rules so that ICD10 variables undergo a slightly
  different sparsity test.


Fixed
^^^^^


* Fixed a bug in the categorical recoding rules for Data Coding `100012
  <https://biobank.ctsu.ox.ac.uk/crystal/coding.cgi?id=100012>`_.



0.19.2 (Friday 26th April 2019)
-------------------------------


Changed
^^^^^^^


* Changes to built-in categories and to ``fmrib`` configuration.


0.19.1 (Thursday 25th April 2019)
---------------------------------


Changed
^^^^^^^


* Changed the default processing rules for ICD10 variables 40001, 40002,
  40006, 41202, and 41204.
* Added ICD10 variables 41201 and 41270 to the default cleaning/processing
  rules.


0.19.0 (Wednesday 24th April 2019)
----------------------------------


Added
^^^^^


* The ``--column`` option now accepts a file which contains a list of column
  names to import.


Changed
^^^^^^^


* The :func:`.icd10.codeToNumeric` and :func:`.icd10.numericToCode` functions
  have been changed to use the integer node IDs in the ICD10 hierarchy
  file. The previous approach could not handle parent categories, nor a small
  number of ICD10 codes which do not have a ``<letter><number>`` structure.
* The :func:`.fileinfo.has_header` function has been made more lenient for
  files with a small number of columns.


0.18.0 (Tuesday 23rd April 2019)
--------------------------------


Added
^^^^^


* New :func:`.icd10.numericToCode` function for converting from a numeric
  ICD10 code representation back to its alphanumeric representation.


Changed
^^^^^^^


* The default binarised ICD10 column name format has been changed from
  ``[variable_id][numeric_code]-[visit].0`` to
  ``[variable_id]-[visit].[numeric_code]``.
* The ``--non_numeric_file`` will not be created if there are not any
  non-numeric columns.
* The built-in ``fmrib`` configuration now includes verbosity and logging
  settings.
* The :func:`.isSparse` function now returns the reason and value for
  columns which fail the sparsity test.



0.17.0 (Monday 22nd April 2019)
-------------------------------


Added
^^^^^


* New ``--non_numeric_file`` option allows non-numeric columns to be saved to
  a separate file (TSV export only).
* Built-in ``fmrib.cfg`` configuration file, which can be used via
  ``-cfg fmrib``.


Changed
^^^^^^^


* The file generated by ``--unknown_vars_file`` now includes variables which
  are known, but are not in an existing category, and do not have any cleaning
  or processing rules specified for them.
* Built-in categories have been updated.


Fixed
^^^^^


* A bug in the column names generated for binarised ICD10 categorical codes
  has been fixed. This bug would potentially have resulted in collisions
  between column names for different ICD10 codes.


0.16.0 (Friday 22nd March 2019)
-------------------------------


Changed
^^^^^^^


* Full variable and datacoding table files no longer need to be provided -
  ``ukbparse`` uses ``ukbparse/data/field.txt`` and
  ``ukbparse/data/encoding.txt`` files, obtained from the UK Biobank showcase
  website, as the basis for recognising variables and data codings. The
  ``--variable_file``/``-vf`` and ``--datacoding_file``/``-df`` options now
  accept partial table definitions - these will be merged with the built-in
  rules (still stored in ``ukbparse/data/variables_*.tsv`` and
  ``ukbparse/data/datacodings_*.tsv``) when ``ukbparse`` is invoked.


Deprecated
^^^^^^^^^^


* The ``ukbparse_htmlparse``, ``ukbparse_join`` , and
  ``ukbparse_compare_tables`` commands.


Removed
^^^^^^^


* The ``--icd10_file`` command-line option has been removed.


0.15.1 (Thursday 21st March 2019)
---------------------------------


Fixed
^^^^^


* Fixed a bug which arose when using the ``--rename_column`` option.


0.15.0 (Monday 18th March 2019)
-------------------------------


Added
^^^^^


* New cleaning function, :func:`.flattenHierarchical`, for use with
  hierarchical variables (e.g. ICD10). The function can be used to replace
  leaf values with parent values.
* New :mod:`.hierarchy` module which contains helper functions and data
  structures for working with hierarchical variables.
* Definitions for all hierarchical UK Biobank variables are located in the
  ``ukbparse/data/hierarchy/`` directory.


Deprecated
^^^^^^^^^^


* The :func:`.readICD10ConfigFile` function has been replaced with the
  :func:`.loadHierarchyFile` function.
* The :class:`.ICD10Hierarchy` class has been replaced with the
  :class:`.Hierarchy` class .


0.14.8 (Monday 18th March 2019)
-------------------------------


Fixed
^^^^^


* Fixed an issue with the :func:`.binariseCategorical` processing function
  being applied to ICD10 codes.


0.14.7 (Sunday 17th March 2019)
-------------------------------


Changed
^^^^^^^


* Changes to default cleaning rules - negative values for integer/categorical
  types are no longer discarded.


0.14.6 (Saturday 16th March 2019)
---------------------------------


Fixed
^^^^^


* Fixed a ``KeyError`` which was occurring during the child-value replacement
  stage for input files which did not have column names of the form
  ``[variable]-[visit].[instance]``.
* Fixed some issues introduced by behavioural changes in the
  ``pandas.HDFStore`` class.


0.14.5 (Thursday 17th January 2019)
-----------------------------------


Fixed
^^^^^


* Implemented a workaround for a `bug <https://bugs.python.org/issue9334>`_ in
  the Python ``argparse`` module.


0.14.4 (Friday 11th January 2019)
---------------------------------


Changed
^^^^^^^


* Updated the default processing rules for variable
  [1120-1150](https://biobank.ctsu.ox.ac.uk/crystal/field.cgi?id=1120).


0.14.3 (Tuesday 8th January 2019)
---------------------------------


Fixed
^^^^^


* Fixed a regression introduced in 0.14.2, where column loading restrictions
  (e.g. ``--variable``) were not being honoured


0.14.2 (Monday 7th January 2019)
--------------------------------


Fixed
^^^^^


* Fixed a regression introduced in 0.14.1, where using the ``--variable`` and
  ``--visit`` options together could cause a crash.


0.14.1 (Monday 7th January 2019)
--------------------------------


Fixed
^^^^^


* If the index columns for each input file have different names, the output
  index column was unnamed.  It is now given the name of the index column in
  the first input file.
* When the ``--column`` and ``--variable`` options were used together, only
  columns which passed both tests were being loaded. Now, columns which pass
  either test are loaded.


0.14.0 (Tuesday 25th December 2018)
-----------------------------------


Added
^^^^^


* New ``--column`` option, allowing columns to be selected by name/name
  pattern.
* ``ukbparse`` can now be installed from `conda-forge
  <https://anaconda.org/conda-forge/ukbparse>`_.


Changed
^^^^^^^


* The index column in the output file no longer defaults to being named
  ``'eid'``. It defaults to the name of the index in the input file, but
  can still be overridden by the ``--output_id_column`` option.


Fixed
^^^^^


* Blank lines are now allowed in configuration files (#2)
* Fix to derived column names for ICD10 variables in default processing rules.


0.13.1 (Thursday 20th December 2018)
------------------------------------


Added
^^^^^


* Unit test to make sure that ``ukbparse`` crashes if given bad input
  arguments.


0.13.0 (Thursday 20th Deember 2018)
-----------------------------------


Added
^^^^^


* New ``--index`` option, allowing the position of the index column in input
  files to be specified.
* The ``--variable``, ``--subject``, and ``--exclude`` options now accept
  comma-separated lists, in addition to IDs, ID ranges, and text files.


Fixed
^^^^^


* Memory usage estimates in log messages were wrong under Linux.


0.12.3 (Tuesday 18th December 2018
----------------------------------


Changed
^^^^^^^


* Changes to new :func:`.fileinfo.has_header` function to improve robustness.


0.12.2 (Monday 17th December 2018)
----------------------------------


Changed
^^^^^^^


* Now using a custom implementation of ``csv.Sniffer.has_header``, as the
  standard library version does not handle some scenarios.


0.12.1 (Saturday 15th December 2018)
------------------------------------


Added
^^^^^


* Added some instructions for generating your own variable and data coding
  tables to the README.


Changed
^^^^^^^


* The ``ukbparse_demo`` script ensures that the Jupyter ``bash_kernel`` is
  installed.
* The ``ukbparse_compare_tables``, ``ukbparse_htmlparse`` and
  ``ukbparse_join`` scripts print some help documentation when called without
  any arguments.
* Added ``lxml`` as a dependency (required by ``beautifulsoup4``).