2.0.0-beta11

What's Changed

• Optimal singular value shrinkage for operator norm by @IlhaH in #492
• [infrastructure] make canopy direct dependency until SomaLogic makes pypi package by @akotlar in #499
• update canopy to somadata by @akotlar in #500
• Add Euclidean and Riemann median calculation functions by @IlhaH in #498

Full Changelog: 2.0.0-beta9...2.0.0-beta11

2.0.0-beta9

What's Changed

• Add nonnegative covariance estimator and tests by @IlhaH in #491
• Add LD maps for hg38 and organize LD maps by build for PRS. by @cristinaetrv in #489
• fix get_user cli and api commands by @akotlar in #494
• [annotation] Remove duplicate entries due to isCanonical by @akotlar in #490
• Optimize PRS C+T preprocess for performance. by @cristinaetrv in #493
• [infrastructure] Improve Python build process by @akotlar in #496

Full Changelog: 2.0.0-beta8...2.0.0-beta9

2.0.0-beta8

What's Changed

• Modified projection method by @austinTalbot7241993 in #473
• [search] Within query slice, paginate by @akotlar in #477
• Incorporated basic shrinkage to adversarial model by @austinTalbot7241993 in #479
• [annotation] Add isCanonical transcript annotation by @akotlar in #478
• Enhance Covariance Matrix Estimation with Symmetry Enforcement and Warnings by @austinTalbot7241993 in #480
• Add shrinkage methods by @IlhaH in #481
• Supervision Augmented PPCA by @austinTalbot7241993 in #485
• Add compressed versions of AD summary stats in hg19 and hg38 by @cristinaetrv in #487
• [search] Improve clnsig analysis by @akotlar in #486

New Contributors

• @IlhaH made their first contribution in #481

Full Changelog: 2.0.0-beta7...2.0.0-beta8

2.0.0-beta7

• Fixes search slicing behavior, so that we are guaranteed to fetch all records, and without having to rely on unbounded number of slices (#477)

2.0.0-beta6

What's Changed

• Soft impute CV by @austinTalbot7241993 in #467
• Feature/ppca regularized adversarial by @austinTalbot7241993 in #469
• Implemented commonly used losses for covariance matrix estimation by @austinTalbot7241993 in #472
• Covariance shrinkage by @austinTalbot7241993 in #471
• Fast dosage matrix filtering by @akotlar in #462
• Feature/no null dosage by @akotlar in #476
• Add back support for nulls in fast dosage matrix calculation by @akotlar in #475

Full Changelog: 2.0.0-beta5...2.0.0-beta6

2.0.0-beta5

• Improve dosage matrix filtering: #462 @akotlar

  • Reduces memory usage during dosage filtering by using B-Tree instead of hashmap to store loci
  • Speeds up sorting loci by sorting each block of 10,000 records fetched in parallel.
  • Reduces memory usage in save handler by writing filtered loci to disk in streaming fashion

• Implement commonly used losses for covariance matrix estimation #472 @austinTalbot7241993

• Ledoit and Wolf covariance shrinkage methods #471 @austinTalbot7241993

2.0.0-beta4

What's Changed

• Competitor SPPCA Methods by @austinTalbot7241993 in #463
• Bugfix marginal likelihood supervision + SVAE by @austinTalbot7241993 in #464
• Covariance Module Improvements by @austinTalbot7241993 in #465
• Feature/streaming proteomics by @dlin30 in #434
• Feature/domain adaptation regularized by @austinTalbot7241993 in #468
• Gaussian mixture fitting of POE by @austinTalbot7241993 in #437
• Improved initialization scheme by @austinTalbot7241993 in #470
• #456: Add SomaScan API support by @akotlar in #466
• Fast dosage matrix filtering in Go by @akotlar in #462

Full Changelog: 2.0.0-beta3...2.0.0-beta4

2.0.0-beta3

What's Changed

• Introduce fast dosage matrix filtering
• Add hg19 version of AD GWAS summary statistics for PRS. by @cristinaetrv in #457
• Implementation Multitask Supervision by @austinTalbot7241993 in #458
• Add preprocess for PRS and C and T PRS calculation. by @cristinaetrv in #414
• GPU Support for PPCA by @austinTalbot7241993 in #461

Full Changelog: 2.0.0-beta2...2.0.0-beta3

2.0.0-beta2

What's Changed

• [search] Fix save handler memory usage by @akotlar in #459

Full Changelog: 2.0.0-beta1...2.0.0-beta2

Bystro 2.0.0-beta1

• Adds OpenSearch 2 support
• Python libraries for ML, Bayesian statistical models, Ancestry score calculations
• Improved annotator with support for nearest gene distance calculations, exact match clinvar, dbSNP, gnomADv4
• Dosage matrix generation
• Merge multiple vcf files (partitioned by locus, row-wise) into one annotation dataset (and combine dosage matrices)
• Initial support for proteomics datasets (TMT, SomaScan)

What's Changed

• Added gitattributes file to preserve LF format on Windows by @mfigurski80 in #57
• [infrastructure] update dependencies, install script by @akotlar in #76
• [config, utils] improve hg19.clean.yml, fetch.pm by @akotlar in #78
• fix based on deployment experience for Cutler lab by @akotlar in #87
• Feature/cythonize by @akotlar in #92
• Bugfix/index asyncio by @akotlar in #93
• Bump openssl from 0.10.46 to 0.10.48 in /search2 by @dependabot in #94
• Bump h2 from 0.3.16 to 0.3.17 in /search2 by @dependabot in #95
• [Issue/96] Fix Messagepack compatibility issues with Docker Perl by @akotlar in #97
• Add contrib and ML package directory for analysis packages in testing phase by @cristinaetrv in #98
• add contrib folder by @poneill in #99
• Create pylint.yml by @akotlar in #100
• [search] Cleanup Python search package, enable by default by @akotlar in #104
• fix python environment on arm64 mac os by @akotlar in #106
• Merge B10 v2 by @akotlar in #105
• Bump crossbeam-deque from 0.7.1 to 0.7.4 in /vcf2 by @dependabot in #112
• Switch pylint to ruff, add pytest dep, break out requirements-dev.txt by @poneill in #118
• [search.save] Issue #116: Remove isal by @akotlar in #119
• Bump grep-cli from 0.1.3 to 0.1.6 in /vcf2 by @dependabot in #113
• Add ancestry record classes by @poneill in #122
• [beanstalkd] Issue #103: Add bystro.beanstalkd.messages tests by @akotlar in #120
• add rough train code by @poneill in #129
• [ancestry] add ancestry infra code by @poneill in #131
• Add logging file to ancestry listener by @poneill in #140
• Bump openssl from 0.10.48 to 0.10.55 in /python by @dependabot in #138
• Run tests during CI by @poneill in #134
• add more generic dummy AncestryResponse to ancestry listener by @poneill in #141
• Fix lingering annotation & utility issues by @akotlar in #145
• Cinch ruff config to strictest ruleset bystro currently satisfies by @poneill in #146
• Bugfix/indexing by @akotlar in #147
• make beanstalkd.worker expect sync functions by @akotlar in #133
• Make ancestry tests eager by @poneill in #149
• Lay groundwork to add ancestry chip model training code by @poneill in #150
• Add code to calculate shared illumina / affymetrix callset by @poneill in #154
• add train_chip_model.py by @poneill in #155
• add s3 upload script for ancestry model products by @poneill in #162
• relax ruff to allow named expressions before returns by @poneill in #158
• Add ancestry listener to beanstalk.clean.yml, startup.yml by @poneill in #174
• remove nonce file for testing branch protection rules by @poneill in #173
• Add CONTRIBUTING.md and docs/style_guide.md by @poneill in #159
• Update lint.yml by @akotlar in #177
• Add s3 handling to pull down ancestry model by @poneill in #176
• enable automatic discovery of tests by pytest by @poneill in #186
• Initialize proteomics module by @poneill in #190
• Remove vcf_dir logic from ancestry listener by @poneill in #191
• Add preprocess 1KGP using gnomad loadings for Ancestry by @cristinaetrv in #184
• fix bug in ancestry listener error reporting by @poneill in #194
• Remove vcf_dir logic from ancestry listener by @poneill in #198
• Add proteomics types, listener by @poneill in #192
• Copy vcf simulator from old branch by @cristinaetrv in #195
• ancestry bugfixes: handle prefixless chromosome names, fix pandas performance warning, make AncestryResponse msgspec compatible by @poneill in #197
• Fix bug in superpop predictions, add tests by @poneill in #208
• Fix random.randint call in generate_random_vcf_index flaky test error by @cristinaetrv in #207
• Fix performance regression in parse_vcf by @poneill in #209
• Describe in CONTRIBUTING.md: 1) forking model, 2) how to checkout and… by @akotlar in #201
• Add gnomad PC loadings into ancestry train by @cristinaetrv in #205
• Add combining gnomad pcs to 1kgp to ancestry code by @cristinaetrv in #210
• Add module for loading / prepping fragpipe tandem mass tag datasets by @poneill in #212
• Feature/tada mixture model by @austinTalbot7241993 in #216
• Supervised ppca by @austinTalbot7241993 in #215
• Feature/tada mixture model ml by @austinTalbot7241993 in #217
• [API] CLI tool by @akotlar in #211
• Feature/ppca by @austinTalbot7241993 in #235
• Update install-perl-libs.sh by @akotlar in #236
• Gaussian Factor Analysis Models by @austinTalbot7241993 in #240
• Add proteomics upload CLI by @poneill in #242
• [ancestry] handle missing data in vcf parsing by @poneill in #265
• Make sure ray starts with serve-dev by @akotlar in #264
• Improve INSTALL.md; remove apt update from install-apt.sh by @akotlar in #239
• Add noqa's to test_ancestry_types.py in order to explicitly check types in ProbabilityInterval by @poneill in #266
• Add loading / prep code for fragpipe DIA datasets by @poneill in #213
• optimize unit test suite by @poneill in #268
• Feature/multivariate tada cleanup by @austinTalbot7241993 in #267
• Feature/dropout ppca by @austinTalbot7241993 in #255
• Create Ticket issue template by @akotlar in #270
• Add caching to lint.yml by @akotlar in #275
• Add project-level config utilities by @poneill in #278
• Add utility functions to get name of GNU tar executable in OS-agnostic manner by @poneill in #276
• Batch API docs by @akotlar in #290
• Add web_api.md, which links to Postman docs by @akotlar in #291
• Perl tidy by @wingolab in #286
• Update Perl Docker recipe by @wingolab in #292
• Update Perl CI by @wingolab in #293
• Perl test cleanup by @wingolab in #303
• Bump rustix from 0.36.9 to 0.36.16 in /python by @dependabot in #301
• Tidy Perl code and docs by @wingolab in #308
• Improve inde...