kissDE R packagekissDE is a R package, similar to DEseq, but which works on pairs of variants, and tests if a variant is enriched in one condition. It has been developped to work easily with KisSplice output. It can also work with a simple table of counts obtained by any other means. It requires at least two replicates per condition and at least two conditions.
Install and use kissDE
In a R interface, first install the package:
Then load the library:
You have now access to the functions of the package. You can refer to the functions documentation like on this example:
Requirements: kissDE package requires several libraries to be installed.
- Note for Ubuntu : DESeq2 installation may fail if the ubuntu package xml2-config is not installed
- For kissDE version 1.3.0 and older, please use DESeq
- aod (available on CRAN)
- xtable (available on CRAN)
- Note: version 2.14 or greater of the DSS package should be used
- glmnet (available on CRAN)
Version 1.4.0kissDE version 1.4.0 (2016-12-01)
- Corrected a bug that prevented the package to work with Biobase version 2.34
- Corrected a bug in dispersion estimation
- Changed dependency on DESeq to dependency on DESeq2
- User's guide
Version 1.3.0kissDE version 1.3.0 (2016-03-16)
- This version corrects a bug that affects the delta PSI/delta f computation. For some genes, the delat PSI/f was not calculated (and set to NA). We advice to re-run kissDE to obtain more candidates with valid delta PSI/f.
- User's guide
Version 1.2.0kissDE version 1.2.0 (2015-09-07)
In this version we fixed some bugs that impact results and we made it possible to deal with more than 2 conditions.
- User's guide
- bugs fixed related to events with low counts/following Poisson law : this may change your results, we highly encourage you to run again kissDE on your data
- >2 conditions testing is effective
- alpha options : kissDE can take as input discoSNP++ files and kissplice2refgenome's files
Version 1.1.0kissDE version 1.1.0 (2015-05-11)
In this new version we improved the recall and made more precise the size of the effect computation.
- size of the effect computation improved
- several bug resolutions
- enables access to pvalues before the correction for multiple testing
Version 1.0.0kissDE version 1.0.0 (2014-10-29)
- Increased speed and sensitivity. More stable with n>2 conditions :
- main function accelerated up to 40%,
- a method to bypass model calculation issues due to counts equal to zero in data has been implemented. As a consequence, p-value results may have changed as multiple correction tests are impacted and more significant events are returned.
- New options:
- pvalue: the user can set his own pvalue threshold,
- pathFigs: the user can give a path where to store the figures,
- filterLowCountsVariants: below this boundary the events are only used the calculate the overdispersion and not to compute the model because they are considered to have too small counts (see the example below),
- flagLowCountsConditions: below this boundary events are flagged in 'LowCounts' column because for at least n-1 condition they have low counts (see the example below).
- New kissplice2counts function:
- makes easier dealing with paired-end data,
- -counts option corresponds to --counts option in KisSplice: values can be 0 (all reads),1 (junctions reads only),2 (separation between junction reads and exonic reads),
- enables to take into account only exonic reads.
- the delta PSI value is now relative,
- the difference of length between the two variants is output instead of the length of the longer variant previously,
- a new column, 'LowCounts' is added to highlight events with low counts in at least n-1 conditions over n.
kissDE version 0.1.0 (2014-05-02)