Changelog

[0.4.0] - 2025-08-24¶

Added¶

• seqspec build subcommand. Generates a draft spec from a natural-language description using the existing init/insert/modify/check tools. Marked experimental.
• New validation checks for seqspec check
  • check_region_against_subregion_length
  • check_region_against_subregion_sequence
  • check_read_length_against_library
• seqspec index:
  • --no-overlap to de-duplicate regions covered by overlapping reads.
  • Target -t kb-single for single-end kallisto-bus indexing.
  • Primer ID may be any region (not only a leaf).
• seqspec print -f png now layers sequencing reads on top of the library diagram.
• load_spec supports gzipped YAML (.yaml.gz).
• seqspec check can skip specific checks; can ignore onlist when needed.
• “Loose loading” mode for seqspec check and seqspec format with captured validation errors.
• New region type: sgrna_target.
• Added private attribute to Assay, _spec_path (str). Stores the fully resolved path to the loaded spec. Used in seqspec check.

Fixed¶

• Read.get_file_by_id returned the wrong file in some cases.

• seqspec file -f paired lists all files.

• seqspec index:

  • Correct file-name indexing with -s file.
  • Correct behavior when targeting region indices.
  • Deterministic deduplication of FASTQ filenames.
  • Lists file URLs and accepts HTTPS onlists.

• seqspec onlist:

  • Local caching and remote download (handles gzip and HTTP 302).
  • Support for kallisto “multi” file format.
  • Fix for bug tracked in #68.

• Plotting fixes for single-modality figures and read overlays.

• Test suite rewritten and passing on modern Python; mocks added for remote I/O.

Changed¶

• seqspec init now initializes an empty seqspec file.
• seqspec creation simplified:
  • Create with init, then add regions/reads via insert, then modify.
  • insert/modify take lists of *Input objects. Consistent -s selector.
  • -r is deprecated in favor of -i across index, onlist, and find.
• Internal models now derive from Pydantic BaseModel. YAML tags (!Assay, !Read, !Region, !Onlist, !File) are no longer required and are stripped on load.
• Region.regions defaults to [] (not optional). Several repr and printing paths were cleaned up.
• Packaging and dev:
  • Moved to pyproject.toml; works well with uv. Versioning via setuptools_scm.
  • Formatter switched to ruff. Test runner switched to pytest.
• seqspec index internals made consistent; primer mapping works at any level of the library.

Removed¶

• spec_fn parameter from seqspec check.
• seqspec convert CLI (unimplemented).
• Legacy implicit parent_id in specs (was unused).
• tox (replaced by pytest).

Breaking changes¶

• seqspec onlist now uses urltype (not location) to decide local vs remote. Older specs must update onlist metadata.
• init no longer accepts library/read structures. You must insert regions and reads after init.
• Input flag change: prefer -i (ID) over -r across index/onlist/find. -r is deprecated and will be removed in a future release.
• Onlist/schema updates since 0.3.x (Files in Reads, file_id fields, URL checks) may require running seqspec upgrade or manual edits.

[0.3.1] - 2024-10-14¶

Added¶

• Added an option to report the full path of the URL of files if the URL type is local.

Fixed¶

• Fixed a bug in the seqspec file with the flag “-f paired” that was causing it to not list out all of the files correctly.

[0.3.0] - 2024-10-10¶

Added¶

• Added typing hints to many Assay and Region functions
• Added 'k key to seqspec info to display one of meta for metadata, sequence_spec for sequence spec, and library_spec for library spec
• Added -f format to seqspec info to enable multiple formats for displaying info
• Added support for 0 length regions in the specification
• Added ability to modify reads using seqspec modify
• seqspec init now accepts read information
• Added templates and updated documentation to myst, website
• Added new examples (e.g., dogmaseq-dig)
• Added File object to Read
• Added SeqKit, SeqProtocol, LibKit, and LibProtocol attributes
• Added option to specify multiple SeqKit, SeqProtocol, LibKit, LibProtocol for each modality
• Added name validation against known list of SeqKit, SeqProtocol, LibKit, LibProtocol
• Added new attributes to Onlist object (Onlist now contains a File object):
  • filetype
  • filesize
  • url
• Added “file_id” to the File object
• Added seqspec file command to return the list of files in the spec as paired across reads or interleaved
• Added seqspec upgrade hidden feature to upgrade v0.2.x seqspec files to v0.3.0.

Changed¶

• Updated seqspec index to read strands for chromap
• Updated seqspec split to correctly split sequence_spec
• Modified seqspec string to use min_len instead of max_len to accommodate nanopore reads
• Updated documentation on how to propose new vocabulary
• Modified the internals of seqspec onlist to manage saving the joined onlist to the -o location when specified (otherwise saves to path where spec lives).
• Renamed “location” to “urltype” in Onlist object
• Replaced -r with -i id type for all functions (deprecated -r)
• Multiple seqspec commands now require the addition of a -s selector [read, region, file]
• seqspec index can now take in reads, regions, or files and index
• seqspec modify can now add files to the template
• Updated splitseq template

Fixed¶

• Fixed seqspec onlist functionality
• Fixed error enumeration in seqspec check

Removed¶

TODO:

• Remove lib_struct
• Remove parent_id

Breaking Changes¶

• File elements are now required in Read objects and Onlist objects in version 0.3.0

[0.2.0] - 2024-04-17¶

Changed¶

• seqspec index uses primer and max length of of supplied Read
• assay_spec renamed library_spec
• Reorganize specification document
• Move contribution guidelines from SPECIFICATION.md to CONTRIBUTION.md
• Move example Regions from SPECIFCATION.md to seqspec/docs/regions
• seqspec index defaults to indexing reads, --region indexes regions
• Change descriptors of attributes assay_id, doi
• Assay attribute assay changed to assay_id
• Read attribute read_name changed to name
• Assay attribute publication_date changed to date
• Assay attribute sequencer changed to sequence_protocol
• Assay function get_modality changed to get_libspec
• Region function update_attr uses the max_len to generate random and onlist sequence lengths instead of min_len
• get_region_by_type changed to get_region_by_region_type to disambiguate between region_type and sequence_type
• seqspec onlist (by default) searches for onlists in the Regions intersected by the Read passed to -r.
• Support older versions of matplotlib by handling the spines[["top", "bottom"...]] structure
• Increase the number of Xs in the random region to match max_len for validation
• Update seqspec print command to use the replacement assay_id attribute instead of assay
• Implement downloading onlists via URLs and transparently decompress gzip files
• Change read_list function to take the onlist object for handling local and remote files
• Added onlist argument to specify combined barcode list file format (kallisto’s multi-file format and default cartesian product format)

Added¶

• Added sequence_spec in the Assay object
• Added Read object in the sequence_spec object
• Added sequence_spec to the seqspec json schema
• Added Read object to specification document
• Added Read generator to website GUI
• Added pattern matching to date in Assay (expected date format: DAY MONTH YEAR, where day is one or two numbers, month is the full named month starting with a Capital letter and year is the full year)
• Added library_kit to Assay object (kit that adds seq adapters)
• Added library_protocol to Assay object (library that generates insert)
• Added sequence_kit to Assay object
• Added website to view example seqspec objects
• Added get_seqspec to assay returns sequence structure for a given modality
• Added multiple checks to seqspec check
  • check read modalities exist in assay modalities
  • check primer ids from seqspec are unique and exists as region ids in libspec
  • check that the primer id exists as an atomic region (currently a strong assumption that may be relaxed in the future)
  • check properties of multiple sequence types
    • fixed and regions not null incompatible
    • joined and regions null incompatible
    • random and regions not null incompatible
    • random must have sequence of all X’s
    • onlist and onlist property null incompatible
  • check that the min len is less than or equal to the max len
  • check that the length of the sequence is between min and max len
  • Note a strong assumption in seqspec print is that the sequence have a length equal to the max_len for visualization purposes
• Added RegionCoordinate object that maps Region min/max lengths to 0-indexed positions
• seqspec onlist searches for onlists in a Region based on --region flag
• Added type annotations for join_onlists to clarify it needs a list of Onlist objects
• Added minimal tests for RegionCoordinate, project_regions_to_coordinates, run_onlist_region, run_onlist_read, and seqspec print functions
• Added list of options to CLI for -f FORMAT within seqspec onlist and seqspec print
• Added -s SEQTYPE to seqspec print to disambiguate printing sequence, library, or libseq objects. TODO wrap seqspec info into seqspec print -f info.
• Added -s SPECOBJECT to seqspec onlist. Specify specific object read, region, or region-type for finding the onlist.
• Added fetching ability for seqspec onlist from remote with IGVF credentials (credit to detrout)

Removed¶

TODO:

• Remove lib_struct
• Remove parent_id

Fixed¶

• Sequencing overlapping pairs now supported
• seqspec check correctly handles sequences lengths longer than the stated min/max range
• Fixed test for project_regions_to_coordinates
• Get the test of seqspec check working again by updating the schema for the refactored example specification YAML files and mocking fastq and barcode files
• Only return the onlist filename if it’s a local file, downloading remote lists when needed