New major versions of kallisto and sleuth have been released. It is recommended that all users upgrade. In particular, a bug that could potentially change the results in sleuth has been fixed. It is recommended that users rerun their analysis with new sleuth.
You can find detailed release notes and binaries here:
---------------------------------------------------
# kallisto
- multithreaded pseudoalignment
- bias parameters and fragment length distribution integrated into HDF5 (integration now exists in sleuth)
- bug fix when read maps to the end of a transcript
---------------------------------------------------
# sleuth
There have been several changes including a very important bug fix. __It is recommended that all users upgrade.__
# new features in the model
- support for likelihood ratio test
- support for arbitrary design matrices
- support for arbitrary normalizations
# bug fixes
- sometimes the samples would get reordered based off of the sample names. This has been fixed.
# miscellaneous
- the interface of `sleuth_prep` has changed to encourage users to correctly specify the connection between samples, kallisto results, and covariates
- imports fragment length distribution from kallisto
- view the bias weights learned from kallisto
- generalize reading from h5 allowing filename or directory to be specified
# visualization and sleuth_live
- most figures and tables can be saved directly from sleuth_live
- fragment length distribution in sleuth_live
- variance explained from principal components can be visualized
- PCA loadings can now be visualized