Releases: SBC-Utrecht/pytom-match-pick
Releases · SBC-Utrecht/pytom-match-pick
0.4.2
0.4.1
0.4.0
What has changed
- The name has changed to pytom-match-pick for easy distinguishing!
- Installation has been changed to use entry-points instead of bin scripts.
- Small fixes for template generation to prevent some nasty artifacts. It can improve results for EM map generated templates.
- Small improvements to the repository to continuous testing.
- Coverage reports are now generated for pull requests!
List of PR's
- Convert bin scripts to entry points by @sroet in #109
- small fixes for template generation and preparation by @McHaillet in #114
- initual guess for github workflow by @sroet in #115
- get gh workflows working by @sroet in #116
- Add test badge to README.md by @sroet in #117
- add coverage reporting by @sroet in #118
- Update coverage.yml by @sroet in #120
- Update unit-tests.yml by @sroet in #121
- Update coverage.yml by @sroet in #122
- test 80% coverage failing by @McHaillet in #123
- and back to 75! by @McHaillet in #124
- enable coverage of the multiprocess code by @sroet in #125
- changing name to pytom-match-pick from pytom-template-matching-gpu by @McHaillet in #126
- add tests for missing libraries by @sroet in #127
- remove unnecessary 'if name==main:unittest.main()' from testfiles by @sroet in #130
- get tests to 100% coverage by @sroet in #131
- readme update pytom-match-pick by @McHaillet in #132
- Update main branch version to 0.4.0 by @sroet in #133
Full Changelog: 0.3.5...0.4.0
0.3.5
Updates
- The calculation of standard deviation over the template matching search has been updated, which means the default extraction mode of
pytom_extract_candidates.py
should work better. This is especially the case for jobs that split the search into subvolumes. - Quality of life improvement, but does not change any results: whitening filter is no longer recalculated everytime the TMJob class is initialized, which saves unnecessary calculation for extraction jobs.
List of PRs
- Slicing for std calculation by @McHaillet in #110
- Prevent whitening for extraction by @McHaillet in #112
Full Changelog: 0.3.4...0.3.5
0.3.4
Major points
- We fixed a bug where in template matching the combination of a restricted search in xyz (
--search-x
option, etc.) and volume split would lead to incorrect merging of subvolumes. - Installation instructions favor building cupy against local cuda-toolkit, but for robustness also highlight the option to put a cuda-toolkit with prebuild cupy in the conda environment. However, building cupy can provide almost 2x better performance on our systems.
- We added a workaround for cupy 13.0.0 which currently broke a custom kernel.
List of PRs
- Improved installation instructions by @thorstenwagner in #69
- Update install instruction p2 by @McHaillet in #71
- Terminate other processes if one breaks by @McHaillet in #75
- parallel manager in a separate test file by @McHaillet in #78
- unittest discover requirest testfiles to start with 'test' by @sroet in #79
- calculate spectrum whitening filter only in the search region by @McHaillet in #81
- Standard deviation cut-off lower threshold by @McHaillet in #82
- Subvolume indexing bug by @McHaillet in #83
- Add scipy to requirements in setup.py by @sroet in #87
- Add tutorial wiki link to the README.md by @sroet in #86
- make json writing OS agnostic in tmjob.py by @sroet in #90
- Template.py: make filter warnings conditional and allow for no ctf_params by @sroet in #98
- experimental tophat filtering for constraining score volume peaks at extraction by @McHaillet in #91
- add docstrings to all functions by @McHaillet in #100
- Convert to pyproject.toml by @sroet in #104
- Cupy13 workaround by @McHaillet in #107
- update install instructions to prefer building cupy by @McHaillet in #108
New Contributors
- @thorstenwagner made their first contribution in #69
Full Changelog: 0.3.3...0.3.4
0.3.3
What's Changed
- update to minimal cutoff value in estimate_roc by @McHaillet in #65
- Patch unittest inconsistency by @McHaillet in #66
- Patch single radial average function by @McHaillet in #67
- Bump version to 0.3.3 by @McHaillet in #68
Full Changelog: 0.3.2...0.3.3
0.3.2
What's Changed
- Patch for failing extraction at volume edges by @McHaillet in #60
- Patch band pass unassigned by @McHaillet in #62
- improve radial average speed with nimdage.mean() by @McHaillet in #63
- add check for equal scores in merging sub jobs to prevent race condition by @McHaillet in #64
Full Changelog: 0.3.1...0.3.2
0.3.1
What's Changed
import pytom_tm
now has apytom_tm.__version__
attribute- jobs are now annotated with the module version number
- the symmetry option is now specific about only covering rotational symmetry around the z-axis
- whitening filters are now stored to prevent recalculation
by @McHaillet in #57
Full Changelog: 0.3.0...0.3.1
Per-tilt CTF and dose weighting, spectrum whitening, and rotational symmetry
Version 0.3.0 is out!
Functionality updates to pytom_match_template.py
:
- The
--per-tilt-weighting
option has been expanded so it can now model a 3D CTF with tilt and dose weighting (as previously described in methods like Relion subtomo averaging, Warp, and EmClarity). To enable the full weighting 3 files need to be provided: a .rawtlt/.tlt file with tilt angle info, a .defocus/.txt file with per tilt defocus (in nm), and a file with accumulated dose per tilt (in e-/A2). The defocus file and dose file are assumed to have the same ordering as the tlt file. PR #53 - Spectrum whitening is now available (similar to Rickgauer et al., (2017) for 2DTM). The power spectrum radial average is estimated for the full tomogram, and the whitening filter is calculated from it and applied to tomogram and template. It sharpens the correlation peaks! PR #53
- There is now a rotational-symmetry option to reduce the angular search for templates with rotational symmetry. PR #55
- It is now possible to provide a custom angular search file. PR #56
Fixes:
pytom_create_template.py
has a more specific warning message for the low pass filter. PR #51
Docs will be updated soon and we are working on a tutorial!
Thanks @rdrighetto for some very useful discussion on the CTF model.
IO consistency fixes and cellular masks
Most important updates
- starfiles written out by this module now contain all the extraction statistics and header start with
_ptm
(#43) - tomographic mask to extract particles within cellular regions can now be applied during candidate extraction (#42)
- a fix has been added to make the testing of parallel jobs numerically stable (#40)
- a few IO fixes for MRC files to make reading more permissive (#33, #38, #41)
Thanks to @hamid13r for the great feedback!