Skip to contents

Changelog

Source: NEWS.md

Version 0.0.20.900

Changes

  • Changed from GPL-2 to MIT license, added proper LICENSE files.

Bug fixes

  • freshenGenes()

    • Fixed empty_rule="original" when ‘ENTREZID’ is populated but not valid, causing the final entries to be empty. Now it properly fills in the original value.
    • Fixed when the intermediate column (‘ENTREZID’) had NA instead of “” for empty values. This situation only occurred when input data.frame also included the intermediate column. Previously these rows were not filled properly.

Version 0.0.19.900

  • Moved ‘AnnotationDbi’ to Imports. ## Bug Fixes

  • Fixed regression from 0.0.18.900 in “first_try” that caused it to behave like “first hit”.

  • Fixed regression with ignore.case=TRUE caused by the correction to handle multiple identical values when converted to lowercase.

  • Added testthat with confirmation of ignore.case=TRUE, revert_split=TRUE, sources, etc.

Changes to existing functions

  • freshenGenes()

    • New argument revert_split=TRUE will re-combine columns previously split by argument split. The process took delimited values, split them into individual columns, then performed matching on the singular values. Previously data were left as-is for review, however the new default is to combine columns using sep as delimiter, for consistency with the delimiter style of this function.
    • New default value for argument intermediate="ENTREZID". Previous 'intermediate' was not useful or informative, although it was flexible. It is still flexible, but the default case is now informative.

Version 0.0.18.900

bug fixes

  • Previously ignore.case=TRUE could in rare cases return a subset of associated genes, only when there were multiple alias values which differed only in case, for example "cal", "Cal", "CAL". This issue only affects values with mixed-case, thus all human entries are unlikely to be affected (most human gene symbols are already UPPERCASE.) Somebody out there recognizes this case, since ‘cal’, ‘Cal’, ‘CAL’ each map 1-to-1 with one Entrez gene, ‘Gopc’, ‘Fblim1’, ‘S100a11’, respectively. Bless.

notable changes to existing functions

  • freshenGenes()

    • When input data contains colname equal to argument intermediate, existing values in that column are considered “complete” and will not have additional values amended alongside them.

Version 0.0.17.900

dependency update

  • Bumped jamba to 1.0.4 and pointed to CRAN instead of Github.

Version 0.0.16.900

dependency update

  • bumped the version dependency on jamba to 0.0.87.900, due to a bug introduced in jamba::cPaste() that adversely affected output for accessions that require multiple rounds of querying.

Version 0.0.15.900

updates to existing functions

  • freshenGenes() new argument ignore.case which calls genejam::imget() as a drop-in replacement for mget(). The process was improved by calling AnnotationDbi::keys() instead of AnnotationDbi::ls(), and this change is at least an order of magnitude faster. In brief benchmarks, using ignore.case=TRUE adds roughly 0.1 seconds per annotation in try_list, but otherwise is the same speed regardless the number of input entries.

Version 0.0.14.900

updates to existing functions

  • freshenGenes() was updated so the input can contain intermediate values, for example ENTREZID values. In fact, now input can contain a mixture of gene symbols, ENTREZID intermediate values, and it will fill in the holes accordingly.

    • new argument intermediate to define the colname that contains the intermediate values, most commonly EG which are Entrez gene ID.
    • Values are propagated in intermediate except when handle_multiple="first_hit" any existing value in intermediate is used with no further processing. All other handle_multiple will combine entries into intermediate.

new functions

  • is_empty() is a small helper function to determine which entries in a vector are either NA or "".

Version 0.0.13.900

new functions

These two new functions are convenience functions. I often find myself wanting the gene symbol and long gene name, so now freshenGenes2() does that by default. To add gene aliases, use freshenGenes3().

Version 0.0.12.900

Changes to existing functions

  • get_anno_db() logic to check for reciprocal annotation names was updated to cover more scenarios. Specifically, "org.Hs.egUNIPROT2EG" is properly recognized, it previously was not being recognized by the reciprocal "org.Hs.egUNIPROT" and therefore was being skipped.

Version 0.0.11.900

Changes to existing functions

  • To prepare for a wider release, I decided to rename (!) some arguments, to have snake_case instead of camelCase for consistency. I heard myself complaining about my own package, “Why are some arguments camelCase and others are snake_case? Pick one!” I complain with a smile on my face, but still it’s a fair point.

    • finalList is now final
    • tryList is now try_list
    • annLib is now ann_lib

I suppose I should probably rename freshenGenes() to freshen_genes().

Changes to existing functions

  • get_anno_db() new argument ignore.case which will build an environment where all keys are converted to lowercase. Ultimately, this option incurs the lowest performance hit, since the keys only need to be converted once, then the environment can be used repeatedly with native mget() functions.

New function

  • imget() case-insensitive mget() – however once I tested it, I realized this mechanism is fairly slow when using a fairly large annotation object. Also, if querying the same data multiple times using imget(), there is no re-use and the cost is incurred each operation – very much not ideal. This function will likely be retired soon.

Version 0.0.10.900

New functions

  • better_exists() and better_get() which are (not so humbly) improved versions of base::exists() and base::get() respectively. Their sole benefit is to recognize a package prefix in an object name, so things like better_exists("base::get") will return TRUE since that object does exist; and subsequently better_get("base::get") will return that object. These functions are mostly useful when using annotation package prefixes such as "KEGG.db::KEGGPATHID2NAME". I needed a simple way to test if it exists. better_exists() also allows multiple input values.

Changes to existing functions

  • get_anno_db() now calls better_exists() and better_get() which allows using a package prefix with annotation names.
  • get_anno_db() argument "revmap_suffix" now allows multiple possible values, it cycles through each until it finds a match, otherwise returns NULL. Some annotations use suffix "2ENTREZID" instead of "2EG", and still others use "2NAME". I’m sure there will be others.

Version 0.0.9.900

Changes to existing functions

  • freshenGenes() new argument handle_multiple="best_each" which returns the best first try for each delimited entry in each input row. For example c("APOE","APOA") will match "APOE" as an authoritatice gene symbol, but "APOA" is matched as an alias to the new gene symbol "LPA". The output will be "APOA,LPA". Note that output will contain unique entries delimited, but they will not be sorted.

Version 0.0.8.900

The next version will have handle_multiple="best_each" which will find the best match for each entry in a set of delimited gene symbols. Most useful for something like pathway enrichment results, where the goal is to retain all possible genes, yet each gene may require a different type of annotation to find a match. See TODO.md for details.

Updates to existing functions

  • freshenGenes() includes a new example showing how to recognize Affymetrix probesets by using a custom search library.
  • freshenGenes() handles multiple annotation libraries, mostly in the form of fully described annotation names, such as "org.Hs.egSYMBOL" and "hgu133plus2ENTREZID".
  • freshenGenes() new option empty_rule="na" which will replace empty entries with NA. Other options empty_rule="blank" replaces with "", and empty_rule="original" replaces empty entries in the first output column with the original entry in the first input column.
  • get_anno_db() now returns`NULL when an annotation is not found, instead of throwing an error. This change allows the calling function to skip missing annotation gracefully without using tryCatch() to catch the error.

Version 0.0.7.900

Updates to existing functions

  • freshenGenes() now properly ignores NA values without throwing an exception. NA values are left as-is and returned as NA in the final output.

Version 0.0.6.900

Updates to existing functions

  • freshenGenes() new argument protect_inline_sep helps to prevent splitting single values that may include the same sep character, for example not splitting "H4 clustered histone 10, pseudogene" into "H4 clustered histone 10" and “pseudogene". Also, the handling of finalList uses sep as the split since that sep is known to have been used in creating the intermediate values, therefore it should be consistent in the final step. This subtle change helps allow a more general split pattern in the first step, such as "[, ]+" which splits at comma and/or space, without splitting at spaces in subsequent steps.

New functions

  • lgsub() is a simple wrapper around gsub() except that it operates on character vectors inside a list. This function will probably be moved into the jamba package in the near future. In fact, there will probably be a small family of list-friendly functions: lgrep(), lvigrep(), lvgrep(), lstrsplit().

Version 0.0.5.900

Updates to existing functions

Version 0.0.4.900

Updates to existing functions

  • Another update to resolve cascade stringsAsFactors bugs. Least favorite default option in R.

Version 0.0.3.900

Updates to existing functions

Version 0.0.2.900

Note that genejam requires one Bioconductor annotation package, usually org.Hs.eg.db but can be any valid organism, such as org.Mm.eg.db for mouse, or org.Rn.eg.db for rat.

Bug fixes