Maintenance

• Support for matplotlib 3.9 (color maps import both from matplotlib.cm and matplotlib.colors)
• Support for numpy 2.0 (explicit np.int32_t in cython functions)

New features

• Add pool decorator to functions for supporting multiprocess
• expected_cis now accepts unbalanced cool file too

API changes

• expected_cis
  • output cvd table now also includes "dist_bp", "contact_frequency", and "n_valid" columns
  • now returns "count.avg.smoothed" and "count.avg.smoothed.agg", when clr_weight_name=None, smooth=True, aggregate_smoothed=True

Maintenance

• Replaced np.int with int in adaptive_coarsegrain
• OE update in sandbox
• Cross score sandbox fixes
• Support for pandas 2

Maintenance

• Bug fix in CLI pileup

New features

• New function/tool rearrange_cooler to reorder/subset/flip regions of the genome in a cooler
• New test dataset for micro-C from hESCs

API changes

• snipping: reorder the axes of the output snipper array to (snippet_idx, i, j).

Maintenance

• snipping: fix spurious nan->0 conversion of bad bins in on-diagonal pileups
• snipping: fix snipping without provided view
• snipping: fix for storing the stack in a file
• virtual4c: fix for the case when viewpoint has no contacts
• fix: Fix numba deprecation warnings by adding nopython=True
• Other small bugfixes

Maintenance

• Updated import statements and requirements to use cooler 0.9.

Maintenance

• Improvements for read_expected_from_file
• Bug fix for dot caller 0/0 occurrences
• Remove cytoolz dependency
• Pin cooler <0.9 until compatibility

API changes

• remove custom bad_bins from expected & eigdecomp #336
• coverage can store total cis counts in the cooler, and sampling can use cis counts #332
• can now calculate coverge for balanced data #385
• new drop_track_na argument for align_track_with_cooler, allows calcultions that that missing data in tracks as absent #360
• multi-thread insulation by chromosome (TODO: by chunk)
• Virtual 4C tool #378

CLI changes

• CLI tool for coverage()

Documentation

• snipping documentation
• dots tutorial
• CLI tutorial

Maintenance

• Dropped support for Python 3.7 (due to Pandas compatability issues)

• Added support for Python 3.10

• Minor bugfixes and compatibility updates

  • Pandas compatibility, pinned above 1.5.1
  • bioframe compatability
  • scikit-learn, pinned above >=1.1.2
  • saddle binedges, value limits #361
  • pileup CLI bugfix for reading features

Other

• Code of conduct

API changes

• cooltools.dots is the new user-facing function for calling dots

Maintenance

• Compatibility with pandas 1.4
• Strict dictinary typing for new numba versions
• Update to bioframe 0.3.3

NOTE: THIS RELEASE BREAKS BACKWARDS COMPATIBILITY!

This release addresses two major issues:

• Integration with bioframe viewframes defined as of bioframe v0.3.
• Synchronization of the CLI and Python API

Additionally, the documentation has been greatly improved and now includes detailed tutorials that show how to use the cooltools API in conjunction with other Open2C libraries. These tutorials are automatically re-built from notebooks copied from https://github.com/open2c/open2c_examples repository.

API changes

• More clear separation of top-level user-facing functions and low-level API.

  • Most standard analyses can be performed using just the user-facing functions which are imported into the top-level namespace. Some of them are new or heavily modified from earlier versions.

    • cooltools.expected_cis and cooltools.expected_trans for average by-diagonal contact frequency in intra-chromosomal data and in inter-chromosomal data, respectively

    • cooltools.eigs_cis and cooltools.eigs_trans for eigenvectors (compartment profiles) of cis and trans data, repectively

    • cooltools.digitize and cooltools.saddle can be used together for creation of 2D summary tables of Hi-C interactions in relation to a digitized genomic track, such as eigenvectors

    • cooltools.insulation for insulation score and annotation of insulating boundaries

    • cooltools.directionality for directionality index

    • cooltools.pileup for average signal at 1D or 2D genomic features, including APA

    • cooltools.coverage for calculation of per-bin sequencing depth

    • cooltools.sample for random downsampling of cooler files

    • For non-standard analyses that require custom algorithms, a lower level API is available under cooltools.api

• Most functions now take an optional view_df argument. A pandas dataframe defining a genomic view (https://bioframe.readthedocs.io/en/latest/guide-technical-notes.html) can be provided to limit the analyses to regions included in the view. If not provided, the analysis is performed on whole chromosomes according to what’s stored in the cooler.

• All functions apart from coverage now take a clr_weight_name argument to specify how the desired balancing weight column is named. Providing a None value allows one to use unbalanced data (except the eigs_cis, eigs_trans methods, since eigendecomposition is only defined for balanced Hi-C data).

• The output of expected-cis function has changed: it now contains region1 and region2 columns (with identical values in case of within-region expected). Additionally, it now allows smoothing of the result to avoid noisy values at long distances (enabled by default and result saved in additional columns of the dataframe)

• The new cooltools.insulation method includes a thresholding step to detect strong boundaries, using either the Li or the Otsu method (from skimage.thresholding), or a fixed float value. The result of thresholding for each window size is stored as a boolean in a new column is_boundary_{window}.

• New subpackage sandbox for experimental codes that are either candidates for merging into cooltools or candidates for removal. No documentation and tests are expected, proceed at your own risk.

• New subpackage lib for auxiliary modules

CLI changes

• CLI tools are renamed with prefixes dropped (e.g. diamond-insulation is now insulation), to align with names of user-facing API functions.
• The CLI tool for expected has been split in two for intra- and inter-chromosomal data (expected-cis and expected-trans, repectively).
• Similarly, the compartment profile calculation is now separate for cis and trans (eigs-cis and eigs-trans).
• New CLI tool cooltools pileup for creation of average features based on Hi-C data. It takes a .bed- or .bedpe-style file to create average on-diagonal or off-diagonal pileups, respectively.

Maintenance

Support for Python 3.6 dropped

NOTE: THIS RELEASE BREAKS BACKWARDS COMPATIBILITY!

This release addresses two major issues:

• Integration with bioframe viewframes defined as of bioframe v0.3.
• Synchronization of the CLI and Python API

Additionally, the documentation has been greatly improved and now includes detailed tutorials that show how to use the cooltools API in conjunction with other Open2C libraries. These tutorials are automatically re-built from notebooks copied from https://github.com/open2c/open2c_examples repository.

API changes

• More clear separation of top-level user-facing functions and low-level API.

  • Most standard analyses can be performed using just the user-facing functions which are imported into the top-level namespace. Some of them are new or heavily modified from earlier versions.

    • cooltools.expected_cis and cooltools.expected_trans for average by-diagonal contact frequency in intra-chromosomal data and in inter-chromosomal data, respectively

    • cooltools.eigs_cis and cooltools.eigs_trans for eigenvectors (compartment profiles) of cis and trans data, repectively

    • cooltools.digitize and cooltools.saddle can be used together for creation of 2D summary tables of Hi-C interactions in relation to a digitized genomic track, such as eigenvectors

    • cooltools.insulation for insulation score and annotation of insulating boundaries

    • cooltools.directionality for directionality index

    • cooltools.pileup for average signal at 1D or 2D genomic features, including APA

    • cooltools.coverage for calculation of per-bin sequencing depth

    • cooltools.sample for random downsampling of cooler files

    • For non-standard analyses that require custom algorithms, a lower level API is available under cooltools.api

• Most functions now take an optional view_df argument. A pandas dataframe defining a genomic view (https://bioframe.readthedocs.io/en/latest/guide-technical-notes.html) can be provided to limit the analyses to regions included in the view. If not provided, the analysis is performed on whole chromosomes according to what’s stored in the cooler.

• All functions apart from coverage now take a clr_weight_name argument to specify how the desired balancing weight column is named. Providing a None value allows one to use unbalanced data (except the eigs_cis, eigs_trans methods, since eigendecomposition is only defined for balanced Hi-C data).

• The output of expected-cis function has changed: it now contains region1 and region2 columns (with identical values in case of within-region expected). Additionally, it now allows smoothing of the result to avoid noisy values at long distances (enabled by default and result saved in additional columns of the dataframe)

• The new cooltools.insulation method includes a thresholding step to detect strong boundaries, using either the Li or the Otsu method (from skimage.thresholding), or a fixed float value. The result of thresholding for each window size is stored as a boolean in a new column is_boundary_{window}.

• New subpackage sandbox for experimental codes that are either candidates for merging into cooltools or candidates for removal. No documentation and tests are expected, proceed at your own risk.

• New subpackage lib for auxiliary modules

CLI changes

• CLI tools are renamed with prefixes dropped (e.g. diamond-insulation is now insulation), to align with names of user-facing API functions.
• The CLI tool for expected has been split in two for intra- and inter-chromosomal data (expected-cis and expected-trans, repectively).
• Similarly, the compartment profile calculation is now separate for cis and trans (eigs-cis and eigs-trans).
• New CLI tool cooltools pileup for creation of average features based on Hi-C data. It takes a .bed- or .bedpe-style file to create average on-diagonal or off-diagonal pileups, respectively.

Maintenance

Support for Python 3.6 dropped