Changelog

Changelog#

v3.12.0#

Release date: 2025-21-04

Added#

A new attribute for all assays: info. This can be used to store arbitrary information of various types including pandas dataframes and dictionaries. Unlike row attributes, column attributes, and layers, it is not confined to any shape. For example, the signature of the clones generated for only the somatic variants would have fewer columns than ids and fewer rows than cells in the assay. Compass stores a lot information in it’s variables, until now there was no easy way of saving it in the h5. These are dataframes with shapes that do not align with that of the assay. These values can now be stored in the assay info as follows:

>>> sample.dna.add_info("somatic_signature", compass.node_genotypes_)
>>> sample.dna.add_info("assignment_probability", compass.probability_)

The assays also now store their palette in the info as a dictionary. sampleinfo() is a shortcut for accessing info for single sample assays.

>>> sample.dna.sampleinfo["palette"]

The functions for handling info are has_info(), add_info(), and del_info()

Updates to COMPASS

  • Update to the latest version of COMPASS

  • Add option to rename compass clones with a new probability value using relabel().

  • Allow passing prefixes for file names to run()

  • Added option to get fractions of each clone using node_fractions()

  • Allow passing directories to run()

  • COMPASS can call LOH clones without calling CNV using the cnv parameter in run().

Plotting updates

  • Cytoband information added to the CNV ploidy signaturemap() and plot_ploidy() figures. These are also shown in the CopyNumber workflow.

  • ticks option to heatmap() to disable plotting of ticks for assay with large number of ids.

  • The scatterplot size modifies depending on the length of the labels to ensure that the plotting area is constant.

Updates to CNV

  • Locally stored gene name annotations are used instead of pulling them from Ensembl when running get_gene_names().

  • When using get_gene_names(), the gene name will always be the best match for the amplicon instead of mutliple values separated by a /

  • get_gene_names() returns the gene names while also adding it to the column attributes.

Updates to DNA

  • get_annotated_ids() returns a list of human-readable names for all the variants in the assay.

  • set_annotated_ids() updates the ids to be human-readable names.

  • dna.Dna.genome() to easily access the genome_version metadata value.

  • Store annotations in the DNA assay info instead of the column attributes to ease their access.

  • dna.Dna.snps() to quickly get the ids that are SNPs.

Functionality updates

  • Option to disable matching of ids in SampleGroup using the match_ids parameter.

  • Ability to load h5 files with only raw counts.

  • whitelist=[] is treated just like whitelist=None in load()

  • Deduplication of barcodes is done using integers instead of sample names and no warning is raised when deduplication of barcodes is performed. It can be manually performed using deduplicate_barcodes() and inverted using normalize_barcodes().

  • Store the palette in the info instead of the metadata of the assay, making saved h5 files valid in format even with the stored palette.

  • sample.Sample.vdj to easily access the VDJ assay in the h5 file.

Fixed#

  • ADO score shown in the figure is the same as the ADO score shown in the table in the VariantSubcloneTable workflow

  • Use the correct genome version from the dna assay in COMPASS

  • get_annotations raises an appropriate error when the genome version is not supported.

v3.7.0#

Release date: 2024-08-05

Added#

Changed#

  • sample.Sample.name is a now a property, and cannot be set. It returns a value according to the current sample_name metadata.

  • assay._Assay.title is a now a property, and cannot be set. It returns a value according to the current sample_name metadata.

  • Behavior of default_label in assay._Assay.set_labels(). When default_label is None, only the labels of the provided barcodes are updated.

  • normalized_counts in compute_ploidy() is no longer used. The read_counts layer is used directly.

  • ANNOTATION_COLUMNS constant was moved to missionbio.annotation.constants

  • Use pynndescent instead of scikit-learn to speed up nearest neighbors calculation during graph-community clustering. Results will not be backwards compatible.

Fixed#

  • Ordering of the barcodes in the heatmap when a subset of the variants are used.

  • Fetching of CNV amplicon gene names for regions where ensembl returns an incomplete response.

  • Allow custom grouping of amplicons for cnv.CNV.heatmap() by passing amplicons to features and x_groups values.

v3.4.0#

Release date: 2024-04-01

Added#

  • Support to pass x_groups to signaturemap() and heatmap().

  • Support to pass variant filters to load().

  • positions(), amplicon_performance(), & panel_uniformity() to quickly get amplicon positions, performance and panel uniformity.

  • Option to hide columns in the variants table of the VariantSubcloneTable workflow.

  • Ability to filter variants through the GUI in the VariantSubcloneTable workflow.

  • override parameter for the heatmap() function which is simply passed to clustered_ids and clustered_barcodes.

  • The first column of the subclone table is frozen.

  • Mandate features when x_groups is provided in heatmap().

  • An appropriate error is raised when any cell has 0 total reads when running NSP.

  • An appropriate error is raised when the annotation API is not available.

Changed#

  • Increased the vertical spacing between the graph and the fishplot from 0 to 0.1.

  • The plotting functions in missionbio.mosaic.plotting were moved to missionbio.plotting

  • missionbio.algorithms.nsp was moved to missionbio.demultiplex.protein.nsp

  • Unpinned scikit-learn and hdbscan as their latest versions are compatible with each other.

  • scikit-learn>1.3.1 is installed by default which results in slightly different NSP calls due to changes to its Gaussian mixture model.

Fixed#

  • Load the whitelist variants correctly when filter_variants=True is passed to load().

  • Nill values of DANN score are shown as empty cells instead of º.

  • name_id_by_pos() does not filter the amplicons.

  • Lineplot in plot_ploidy() does not connect the medians with a line when using genes+amplicons or positions+amplicons.

  • The violin plot range is fixed to (0, 100) for the AF and GQ layers in VariantSubcloneTable.

  • Violin plots generated using violinplot() are equally spaced when split by labels.

  • Fix resetting of selected_bars when scatterplots are created.

  • rename_labels() allows swapping of labels.

  • Fishplot does not disappear when a clone and its parent both have 0 cells at some timepoint.

v3.1.1#

Release date: 2023-09-25

Added#

  • Relaxed missionbio.h5 requirement to >=4.13.0,<6

Changed#

  • Disable autouploading of tagged packages to anaconda.

  • Removed check for h5 file compatibility with H5Reader.

Fixed#

  • The whitelist option in load() correctly loads exact matches of variants.

v3.1.0#

Release date: 2023-09-13

Added#

  • The order of the names in the legend matches the order of the traces in the ridgeplot.

  • Option to pass any sequence type to get_attribute() besides np.ndarray. This includes list, tuple, and range.

  • features parameter to signature() which allows grouping across ids, just like splitby allows grouping across cells. * The feautures option in signaturemap() allows plotting using grouped data from signature()

  • Support for hg38 along with all species available through Ensembl in get_annotations()

  • Support for hg38 in get_annotations().

  • Sped up NSP by 2x by using statsmodels for the KDE and using spherical covariance with kmeans++ initialization for the GMM parameters.

  • ANSP - Approximate NSP to protein normalization. It runs in constant time for large datasets.

  • get_attribute() also accepts dataframes.

  • heatmap() can plot arbitrary dataframes as long as it has the expected number of cells.

  • TreeGraph now supports html tags like <br>, <b>, and <span> in the descriptions.

Changed#

  • Use latest python 3.8 in installer instead of 3.8.0

Fixed#

  • The title of clone_vs_analyte() plot does not overlap with the DNA heatmap.

  • The x-axis label order for CNV in the clone_vs_analyte() plot matches the order of the points in the data shown.

  • NGT layer not modified after running filter_variants()

  • “Last modified” timestamp does not change when loading an H5 file.

  • jitter parameter in NSP works

  • Failure of VariantSubcloneTable when all the variant calls are filtered.

  • Pinned hdbscan to v0.8.29. Higher versions (>=0.8.30,<=0.8.33) have runtime issues.

  • heatmap() and signaturemap() execute successfully when “cnv” is passed before “dna”.

  • Fix y-compression of TreeGraph by checking the upwards and downwards movement of only the highest and lowest nodes respectively.

Updated#

  • Switched from using the depracated JupyterDash to the builtin jupyter dash in Dash v2.11. Documentation

  • jupyter_client from <8 to >=8.1.0 as the ThreadedZMQStream error is fixed in it. Changelog

v3.0.1#

Release date: 2023-06-20

Added#

Changed#

  • Updated matplotlib dependency from <=3.2.2 to >=3.4.0

Fixed#

v3.0.0#

Release date: 2023-06-16

Added#

  • A wrapper for COMPASS.

  • New variant filters that account for missing data.

  • Recipe and instructions for building installers.

  • plot_kind parameter to dna.Dna.group_by_genotype() to change the type of plot shown.

  • filter_cells to io.load() which loads only the intersection algorithm cells.

  • Progress bar to io.load()

  • algorithms.nsp.NSP and algorithms.nsp.ExpressionProfile to modularize the NSP code.

  • x_groups to assay._Assay.heatmap() to group the x-axis by a given list of ids.

  • Simplify and speedup assay._Assay.heatmap() by removing duplicate data. (By using plots.heatmap.Heatmap)

  • assay._Assay.convolve() to convolve the data that was earlier performed in the Heatmap.

  • Configuration options accessible via Config:

    • ms.Config.Colorscale.Dna to change the default color palette for all DNA plots.

    • ms.Config.Colorscale.Cnv to change the default color palette for all CNV plots.

    • ms.Config.Colorscale.Protein to change the default color palette for all Protein plots.

  • Custom divirgent colorscale for Cnv Ploidy heatmaps

  • Option to return indices instead of barcodes in assay._Asasy.clustered_barcodes().

  • sample.Sample.common_barcodes() to get the common barcodes across assays.

  • Add subcluster paramter to assay._Assay.clustered_barcodes() to prevent clustering within the labels

  • Option to pass n-dimensional arrays as splitby in assay._Assay.clustered_barcodes()

  • Option to fetch a subset of the assays in sample.Sample.assays() using the names parameter

  • sample.Sample.clustered_barcodes() to hierarchically cluster using multiple assays

  • Multiple options added to sample.Sample.heatmap() to sort the assays, barcodes, and the features

  • assay._Assay.signature`() accepts a splitby parameter to get the signature for each unique label in splitby.

  • Improvements to assay._Assay.signaturemap():

  • Option to copy the labels and palette together by passing an assay._Assay() to assay._Assay.set_labels()

  • assay._Assay.heatmap() sets subcluster=False when calculating the barcode order when convolve is provided.

  • Varsome URLs as hyperlinks on the variant name in the VariantSubcloneTable

  • Add percentage of cells and amplicons present to the CopyNumberWorkflow

  • dna.Dna.mutated_cells() to get the number of cells with at least 1 mutation in each given clone. This is used in sample.Sample.signaturemap().

Changed#

  • apply_filter changed to filter_variants in io.load()

  • SubcloneTree and SubcloneTreeGraph classes are renamed to Tree and TreeGraph respectively.

  • show_plot to return_plot in dna.Dna.group_by_genotype()

  • plots.heatmap.Heatmap splits the vertical and horizontal lines on the main heatmap into two traces.

  • The default value of vaf_het in dna.Dna.filter_variants() changed from 35 to 30.

  • Flattened sample.Sample.heatmap`() option has been removed. A more customizable version is available under the sample.Sample.signaturemap() function.

  • The constant - constants.COLORS to have unique values.

    • The grey values at the 10th, 20th, 30th.. positions were modified to be unique

    • The black (#000000) value was moved from the 20th position to the last position

Fixed#

  • Get indexes maintains the order as per find_list when there are duplicates in the find_list and order_using_find_list is True.

  • DANN score in the variants subclone table is shown correctly for saved h5 files.

  • Overlapping of text in phylogeny trees.

  • Error in multiprocessing when fetching gene_names for CNV by adding a max_workers parameter and using threads instead of processes.

  • Missing clone is ignored when finding ADO sisters.

Removed#

  • Functions to convert legacy loom files to h5 files - io._loom_to_h5, io._update_file

  • Functions to read data from csv files - io._merge_files, io._cnv_raw_counts, io._protein_raw_counts

  • Function to merge h5 files - io._merge

  • show_plot from protein.Protein.normalize_reads(). The same plot can be created in plotly using algorithms.nsp.NSP.plot()

  • show_plot from protein.Protein.get_signal_profile(). The same plot can be created in plotly using algorithms.nsp.ExpressionProfile.plot()

  • protein.Protein.get_signal_profile function. It can be executed using algorithms.nsp.ExpressionProfile.fit() if needed.

  • protein.Protein.get_scaling_factor function. It can be executed using algorithms.nsp.NSP.scaling_factor() if needed

Contents