Skip to content

Latest commit

 

History

History
154 lines (114 loc) · 5.96 KB

CHANGELOG.md

File metadata and controls

154 lines (114 loc) · 5.96 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

1.4.1 - 2024-08-01

Fixed

  • Added packaging to environment.yml
  • Cython extension compiled in C99 mode
  • Fix overflow when initializing random number generator

1.4 - 2023-08-19

Added

  • Updated README.md
  • Tutorial

1.4.0.rc1 - 2023-08-19

Fixed

  • Error when formatting elapsed times longer than 1 day

1.4.0.rc0 - 2023-08-16

Added

  • Implemented stages. Workflows for stellarscope assign and stellarscope resume are refactored into stages. Future plans to implement separate subcommands for each stage.
  • Model selection. Stellarscope reports Bayesian information criterion (BIC) and Akaike information criterion (AIC) for model selection.
  • Output statistics as TSV. Statistics that are reported in the log are also output to stats.final.tsv for easier downstream parsing.
  • UMI deduplication graph visualization. Included an example jupyter notebook for parsing the UMI tracking file and visualizing UMI graphs.

Changed

  • Stranded mode annotation data structure. Previous implementation used the same annotation for stranded and unstranded mode where an interval tree is used for each chromosome and strand match is determined after searching the interval tree. New implementation (for stranded) uses two interval trees for each chromosome - one for each strand - meaning the total height of each interval tree should be about half, assuming no strand bias for features.
  • Parallel barcode aggregation. Counts for cell barcodes can be aggregated in parallel using --nproc option.
  • Renamed --whitelist to --filtered_bc. The --whitelist option is still available for backwards compatibility.

Fixed

  • Verify samtools version. stellarscope cellsort requires that samtools view has the -D option to provide a tag file. This was not present in samtools v1.8, pinning v1.16.
  • Miscellaneous floating point errors.

Deprecated

  • Raising deprecation warnings on various methods of Telescope objects.

Removed

  • --devmode option.
  • Telescope class
  • Stellarscope.output_report_old() Early version of output_report that does not use multimapper UMI correction
  • Stellarscope.print_summary()

1.3.3.dev3 - 2023-06-16

Added

  • Implemented checkpointing of Stellarscope run. Stellarscope object is serialized using pickle. Two checkpoints are currently implemented, one after loading alignment and a second checkpoint after UMI deduplication.
  • stellarscope resume enables resuming from checkpoint file. Users may choose a different pooling mode from the initial run, or specify an alternate celltype file.
  • UMI deduplication summary. Reports number of duplicated UMIs and number of components found per UMI.
  • Improve reported information. Information output to log is more clear and important information is included. Implemented statistics module with FitInfo, PoolInfo, etc. for tracking and displaying summary information.

Changed

  • Consistent output for "random" reassignment modes when using stellarscope resume. Seed for random number generator is stored in the checkpoint object and transferred when resuming.
  • Improved performance of UMI deduplication. Removed asserts and sanity checks. Implemented shortcut for selecting UMI representatives - building adjacency matrix is avoided if all reads share 1 or more features.
  • Parallel model fitting. Pooling modes that fit multiple models (individual and celltype) can perform fitting using parallel processes using --nproc option.

Fixed

  • Miscellaneous floating point errors
  • Pysam import error

1.3.3.dev2 - 2023-01-18

Added

  • CHANGELOG.md was added to project

NOTE: this version was created to comply with pip version requirements, which did not like "1.3.3-fix".

1.3.3-fix - 2022-12-16

Added

  • Implemented checkpointing of Stellarscope run. Stellarscope object is serialized using pickle. Two checkpoints are currently implemented, one after loading alignment and a second checkpoint after UMI deduplication.
  • stellarscope resume enables resuming from checkpoint file. Users may choose a different pooling mode from the initial run., or specify an alternate celltype file.

Fixed

  • Fixed floating point errors. These are typically underflow errors when calculating very small probabilities. Switched to extended precision when encountered.

1.3.3 - 2022-09-12

Fixed

  • Fixed bug where values calculated using the "old" TelescopeLikelihood object were being used to generate the final count matrix. Any results from prior versions should be rerun.

Deprecated

  • fit_telescope_model() is no longer called by default. May be enabled with --old_report. This implies that:
    • "Old" reports (i.e. TE_counts_old.mtx, ...) are no longer generated
    • Run stats are currently not being output in a file.
    • Post-EM UMI duplication is deprecated (UMIs are deduplicated before EM)

[1.3.2] - 2022-09-09

[1.3.1] - 2022-09-06

[1.3] - 2022-09-02

[1.2.1] - 2022-08-15

[1.2.0] - 2022-07-26

[1.1.0] - 2019-02-15

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.