332 lines (227 loc) · 8.47 KB

Version log

Version 1.10.2/1.11.2

Added helper functions to compute the exact moments, so that the user can characterise the systematic biases in the diffusion scores

Version 1.10.1/1.11.1

Fixed issue in Rcpp code, due to deprecation in arma, info here
Updated some warnings and notes from BiocCheck()
Updated readme to link to bioconductor and the journal publications

Version 1.3.1

Fixed vignette title and citation

Version 1.3.1

Fixed vignette title and citation
Small changes to documentation
Version number not malformed anymore

Version 1.0.0

Overall summary:

Five diffusion kernels available, they can be computed from an 'igraph' object.
Diffusion implementations divided between 'diffuse_raw' for deterministic scores and 'diffuse_mc' for permutation analysis, which is parallelised. In total, seven diffusion scores are accessible through the 'diffuse' function.
Performance evaluation wrapped in the 'perf' function.
Helper functions in helpers.R (to plot diffusion scores, to check if a kernel matrix is actually a kernel, to extract largest CC from a graph)

Version 0.103.4

Fix bug

Version 0.103.3

Fixing warnings in doc

Version 0.103.2

Reversed changes in Rcpp

Version 0.103.1

Changed defunct latex2 to latex
Fixed links in doc files
Checks on graph: must be undirected

Version 0.101.1

Added CITATION to the manuscript

Version 0.101.0

This version includes the requests from the reviewers

Documentation is more complete now, especially in diffuse.
- Examples on more scores, not only raw
- Illustration on binary/continuous, differences in unlabelled/negatives
- Definition of the different scores not only in the vignette, but (briefly) in ?diffuse
- Description of kernels in ?kernels
- References to the original publications
Added the following functions, along with documentation, examples and unit testing:
- perf_wilcox: test metrics through Wilcox test
- diffuse_grid: sweep parameters to compute diffusion scores
- metric_auc, metric_fun: provide basic performance metrics
Switched dependency from Metrics to precrec
Added reshape2 as a dependency in Suggests, for its usefulness in cast
Changes in main vignette, to comply with former points
Small update on the main figure in the vignette

Version 0.99.12

Small changes in vignette
Pushing to see if bot notices

Version 0.99.11

No changes

Version 0.99.10

Modified diffuse_mc so that it accepts quantitative inputs (but still should better be sparse)
Scores ml and gm now throw an error if fed with quantitative inputs
This version will be proposed to Bioconductor

Version 0.99.9

This version addresses the first reviews from Bioconductor

Changed the vignette name from yeast.Rnw to diffuStats.Rnw
Fixed some references in vignette
Remove commented out code lines
Remove calls to gc()
Remove commented calls to browser()
Remove Cpp commented code
Vectorised code in diffuse.R (gm method)

Version 0.99.8

Fixed check NOTE in vignette

Version 0.99.7

Same as 0.99.4

Version 0.99.6

Same as 0.99.4

Version 0.99.5

Same as 0.99.4

Version 0.99.4

Changed the package name to diffuStats
More detailed package description in the DESCRIPTION file

Version 0.99.3

Same as 0.99.1

Version 0.99.2

Same as 0.99.1

Version 0.99.1

Fixed eps that was a symlink
Main figure now with the right size

Version 0.99.0

Changes in both vignettes, using Bioconductor style
Submitting to Bioconductor

Version 0.97.0

Changes in main vignette
Ready to submit to Bioconductor

Version 0.5.5

Final adjustments in main vignette
Added example to quickstart vignette
First internal review version

Version 0.5.4

Changes in vignette
Temporarily depends on R>=3.2 to build a zip for iPsych

Version 0.5.3

Modified main vignette: more proper introduction, more cites
Added code chunk to generate the figure in the paper

Version 0.5.2

Added basic sanity checking
Fixed small bug: ml could be wrongly codified
Sanity also checks the variance in the input - useful to detect pathological cases
Reduced n.perm in vignettes and examples - they should build faster
Added unit testing for sanity checkers

Version 0.5.1

Vignette has been added a "proper" introduction
Now main vignette is in sweave

Version 0.5.0

Fixed devel version numbering x.y.z, y must be odd
Added correct indentation, built indentation-fixing function for Rd files
Fixed CRAN's NOTE on R 3.4 about the c++ code
Added NEWS file

Version 0.4.3

Fixed small bug in gm
Added unit testing for inputs using kernels

TODO

Some more unit testing (helpers...). Aim at 100% coverage
Test installation on other systems
Sanity check for inputs
Optional: one kernel wrapper with abbreviations

Version 0.4.2

Added examples to all functions. BiocCheck not complaining anymore!
Fixed small bugs

TODO

Some more unit testing (helpers...). Aim at 100% coverage
Test installation on other systems
Sanity check for inputs
Optional: one kernel wrapper with abbreviations

Version 0.4.0

Added all the kernels and respective helpers, unit testing.
diffuse, perf now accept kernels as inputs as well. Also, the default Laplacian kernel is computed using kernel functions

TODO

Some more unit testing (helpers...). Aim at 100% coverage
Add examples to each function
Test installation on other systems
Sanity check for inputs
Optional: one kernel wrapper with abbreviations

Version 0.2.0

Added performance wrapper: perf (and its core perf_eval) Can apply diffusion over a grid of parameters and compute a metric using "target scores". Unit testing written as well.
Dropped pROC dependency, added Metrics
Helper functions to deal with the format of the files

TODO

Some more unit testing (helpers...). Aim at 100% coverage
Add kernels
Add examples to each function
Test installation on other systems
Sanity check for inputs

Version 0.0.10

Added first 'wrapper': input scores can be given as vector, matrix or list of matrices. Whatever format was given will be returned. Contains unit testing. This simplified vignettes a bit as well.
Modified graph_toy to contain examples of input vector and matrix. Updated doc and first vignette.
Dense Matrix objects converted to base matrix in diffusion output.

TODO

Exhaustive unit testing
Add kernels
Add examples to each function
Test installation on other systems
Add wrappers for several scores screening, AUC, CV(?)

Version 0.0.8

Added largest_cc helper
First version of the yeast vignette. Takes 1:15 to build.

TODO

Exhaustive unit testing
Add kernels
Add examples to each function
Test installation on other systems
Should we add wrappers? (cross-validation scores, AUC... We can use try to give NA as an AUC value for bad data splits)

Version 0.0.6

Fixed bug in ber_s (both matrices are from Matrix)
Fixed bug in gm (missed drop = FALSE for the single-column case)
Started second vignette with a bio example: yeast ppi. Added suggests.
Now using igraphdata, the example is perfect!
raw and ber_s will coincide if the tested nodes do not belong to the background

TODO

Exhaustive unit testing
Add kernels
End vignette

Version 0.0.4

Added synthetic input generator
Fixed a bug in z-score - sums were computed over the raw scores and not the input
Changed internal representation of default values for graph generation
Added more unit testing (still missing some)
Started synthetic example, see data-raw

TODO

Exhaustive unit testing
Add kernels
Add biological example - search for a Bioconductor package with a network in igraph format
Add example
Investigate: are raw and ber_s the same? They should be similar but not the same

Version 0.0.2

First version of the package. Includes:

Synthetic network generation (igraph backend)
Diffusion scores: raw, ml, gm, z, mc, ber_s, ber_p
Monte carlo trials are in C++ and parallelised
Support for diffusion in batchs with different backgrounds
Unit testing of the majority of functions
Example data
Vignette
Helper functions to plot colours and shapes

Results of devtools::check(): 1 NOTE (line SystemRequirements: GNU make in DESCRIPTION file)