Skip to content

Latest commit

 

History

History
84 lines (45 loc) · 3.32 KB

VERSIONS.md

File metadata and controls

84 lines (45 loc) · 3.32 KB
Raw

scrattch.mapping v0.52.2

Major change to how we return mapping results.

Major changes

taxonomy_mapping now returns an S4 class called mappingClass. We've included a single method getMappingResults() to extract the results table from the S4 class.

Minor changes

Many bug fixes and path handling

--

scrattch.mapping v0.4

Remove code associated with taxonomy building to scrattch.taxonomy.

Major changes

Minor changes

--

scrattch.mapping v0.21

This version of scrattch.mapping additionally simplifies the PatchSeq workflow and file structure by moving content from additional files to the h5ad and by allowing for a precalculated dendogram to be input to buildTaxonomy().

Major changes

  • buildTaxonomy() allows for a precalculated dendrogram to be input (back-compatible, so by default the dendrogram is generated).

  • Move the patchseqQC variables from file to Anndata object (only only remaining files outside h5ad are dend.RData, de.genes.rda, and membership_information_reference.rda, which we plan to move to the h5ad in future updates).

Minor changes

  • Create a new example for generating the Hodge et al 2019 MTG taxonomy and mapping patch-seq to it.

  • Allow loadTaxonomy() to read either a folder location or an anndata file directly.

  • Error correction in addDendrogramMarkers (and other functions).

  • Auto-correct '' --> '/' for different platforms in file paths to allow compatibility betwee Windows and Unix, through use of a file.path() wrapper (beta).

scrattch.mapping v0.2

Updates to this version of scrattch.mapping are primarily associated with improving and simplifying the PatchSeq workflow by allowing for Taxonomy versions which filter out target cell types and can be re-used once built. Moving towards removal of label.cols from taxonomy_mapping().

Updated examples for new workflows are provided on our github page.

Major changes

  • buildPatchseqTaxonomy() saves off-target cells into the taxonomy anndata for filtering to specific cell types. To be generalized past patchseq in the future.

  • Enabled the dynamic modification of a base taxonomy into a filtered taxonomy version for mapping using Corr, Tree and Seurat.

  • Taxonomy versions are stored in the .h5ad file and can be reused in future mappings by an user.

Minor changes

  • buildTaxonomy() now also creates the anndata object, this was done only in loadTaxonomy previously.

  • New parameter to buildTaxonomy(), taxonomyName allows the user to encode a taxonomy ID that is stored in the taxonomy .h5ad which is currently fixed at AI_taxonomy.h5ad.

  • loadTaxonomy() will use the anndata object if it exists to load the taxonomy. If not a new anndata object will be created from the feather files and saved while loading.

  • Exposed tree mapping parameters, can now set p, low.th and bootstrap now.

  • Ensure that the dendrogram labels ("cluster_label") is used for all cluster level objects instead of "cluster_1, cluster_2 ...".

  • Removed all dependence on "reference.rda", which is also no longer created.

  • Move multiple functions to internal only, not exposed to the user.

Parameter renaming

(new) | (old) taxonomyDir | shinyFolder taxonomyDir | refFolder

Function renaming

(new) | (old) buildPatchseqTaxonomy | writePatchseqQCmarkers