Pyseer Versions Save

v1.3.11 (May 2023)

• Improvement: Rtab files can use a empty string to signify missing values
• Bugfix: avoid crashes wn running wgGWAS with covariates
• Bugfix (summarise_annotations.py): more robust parsing
• Bugfix: avoid crashes when distance matrices are not squared
• Bugfix: properly format output when using --no-distances

v1.3.10 (July 2022)

• Docs: more clarity on effect sizes for LMM
• Bugfix (similarity): make sure the script actually uses all variants
• Bugfix (QQ-plot): avoid axis inversion with newer versions of statsmodels (thanks to Julian Libiseller-Egger)
• Bugfix: use kwargs when invoking functions from scikit-learn
• Bugfix (annotate hits): avoid crashes when full path to reference genomes contains non-ASCII chars
• Bugfix: WG models should not be ran with the --output-patterns function
• Bugfix: avoid a crash when saving a WG model with covariates
• Bugfix: add the "lineage" header when running a whole genome model

v1.3.9 (June 2021)

• Bugfix: avoid a crash when providing lineages in whole genome mode

v1.3.8 (May 2020)

• Improvement: fall back to Firth regression when encountering a matrix inversion error (thanks to Julian Libiseller-Egger)
• Bugfix: check for zero passing variants in read_all (enet, thanks to Julian Libiseller-Egger)
• Bugfix: use len(all_strains) instead of len(sample_order) to determine shape of sparse matrix in load_all_vars (enet, thanks to Julian Libiseller-Egger)
• Bugfix: properly report all filtered variants
• Bugfix: --lmm requires either --similarity or a LMM cache

v1.3.7 (May 2020)

• Enhancement: check that provided phenotypes are of numeric type
• Bugfix: properly report all filtered variants
• Bugfix: don't crash if regression fails for missing data
• Bugfix for whole genome regression and lineages
• Bugfix/docs: properly report that the covariates file should have a header
• scripts/summarise_annotations.py can work with unadjusted p-values (new option, thanks to Lindsay Clark)
• scripts/phylogeny_distance.py: no need to reroot twice with updated dendropy
• CI: transition to GitHub actions

v1.3.6 (May 2020)

• Bugfix for missing variants in VCF files (now properly handled)
• Bugfixes for k-mer mapping (lack of annotation, bwa fastmap with many hits
• Explicitly look for duplicated samples in input files
• Removed package from pypi (glmnet_py dependency could not be installed that way)

An important fix for generating .plot files, which will be out-of-order and therefore wrong if supplementary mappings are reported (which I think is more likely with unitigs)