Releases: cfe-lab/MiCall
v7.17.1
New Contig Stitching Algorithm
The updated contig stitching algorithm in MiCall now stitches contigs directly, avoiding reliance on reference genome regions, which improves accuracy. Both stitched and unstitched files support workflows like the proviral pipeline. Diagnostic tools include SVG visualizer plots and log files for transparency. More details at issue 1030.
Update to HIVdb v9.4
MiCall now uses HIVdb v9.4
, enhancing resistance interpretation. Key updates include score changes for NRTI drugs (ABC, AZT, TDF, 3TC/FTC) and revised resistance profiles for Doravirine. More details at issue 884.
Simplified Coverage Plots
Improvements to coverage plots include the removal of V3 overlaps, external legends, and visual markers for mutations, reducing clutter and enhancing readability. More details at issue 537.
Fixes to Concordance Plots
Errors in concordance plots have been fixed, addressing issues related to contigs ending with large insertions, alignment corrections, and small alignment errors. Details in issues 887, 963, and 899.
Updates to proviral Pipeline Integration
Enhancements include HIVIntact backend support, automatic downloading of proviral_landscape.csv
, and suspension of proviral uploads to QAI to prevent failures. More details in issues 994, 984, and 886.
New Project Code: HIV-AD
Introduced HIV-AD
, modeled after the HIV-GHA
project, involving cloning the project code, updating PCR primers, and QAI implementation. Details at issue 998.
Code Quality Improvements
Enhancements include:
- Linting Support: Added flake8 for code quality (pull 1055).
- Type Annotations: Introduced mypy for early error detection (pull 1035).
- Dependency Updates: Updated major dependencies.
- OS Update: Moved to Debian 12 for Singularity images (pull 1018).
- New Tools and Features: Added
aln2counts_simplify.py
1 and streamlined aligntools (pull 1023, 1068). - Floating Point Handling: Improved
update_qai.py
2 reliability (pull 1075). - Split Singularity Test: Separated Singularity test in CI (pull 1080).
Explore the GitHub milestone for an overview: milestone.
Additional resources:
HIVdb 9.4
Stitched consensus
The main changes were:
- Added a stitched consensus #745
- Updated to HIVdb 9.0 #686
- Noncoding regions #589
- Launch proviral pipeline #706
- Added chimeric Ghana references for HIVGHA project code #806
- Added concordance evaluation #821
For the full list of issues, see the milestone.
Note: v7.15.12 is only a change to the MiCall watcher to suspend the proviral upload into QAI, as it is currently failing.
v7.15.13 pins merge-mates to the commit that I used to test v7.15 (there have since been new commits that break our code)
To do when putting the new release in place:
- Add Charlotte to email recipients
- Add a new sample to the list of samples to test: 90906A-HCV_S122 from run 16-Oct-2020.M01841
Primers by project
SARS-CoV-2 support
De novo assembly
The big change is #442, adding a de novo assembly method to the standard remapping method.
Also included:
- #486 a project region for GP120, and report V3LOOP coverage with G2P alignments while GP120 reports coverage from remapped or assembled reads.
- #524 Insertion counts are wrong.
- #471 and #473 improve efficiency of micall_watcher script.
See the milestone for a complete list.
De novo assembly (preview)
Development preview of #442 de novo assembly.
HCV subtype bug
Container Runs
The main improvements in this release are in the MiCall watcher, specifically changing to Kive's new container runs (#460). Related fixes are #458 and #459.
- Update HIVdb to 8.8 (#448).
- Some bugs when combining MIDI and whole genome samples: #469, #468, #465, #463, #462, #454, and #450.
For the complete list of issues, see the milestone.
Configuration changes:
- Convert old Kive runs to container runs.
- Upload pipeline definitions to QAI for #441.
- Remove bowtie2 from Kive's host system, because it interferes with the version inside the Singularity container.
- Update kiveapi to 0.12.
HCV Reports
Add HCV reports (#412), then update HCV resistance rules (#436).
- Combine whole-genome and MidHCV samples to get complete coverage in HCV reports (#427 and #438).
- Fix a couple of bugs in the consensus sequences (#417 and #420).
- Add more details to resistance results (#445).
- Fix external datasets (#452).
- Error handling in micall_watcher (#449, #451, and #453).
For the full list of changes, see the milestone.