pcaNet-package: Probabilistic principal components analysis - covariance...

In HGray384/pcaNet: Probabilistic principal components analysis - covariance estimation and network reconstruction

Description

Various implementations of algorithms for probabilistic PCA, with an emphasis on covariance matrix estimation and network reconstruction in the presence of missing values.

Details

Algorithms for PPCA have been ported from the PCAMV MATLAB toolbox (Ilin and Raiko, 2010) and extended from the pcaMethods (Stacklies et. al., 2007) R-package to focus on covariance matrix estimation and network reconstruction in the presence of missing values. Full PCA functionality with pcaMethods is retained in pcaNet due to the use of the pcaRes class.

The inverse of the covariance matrix from PPCA can be computed efficiently, and this functionality is provided in ppca2Covinv. Using the false discovery rate method from Strimmer (2008), the estimated partial correlations can be tested to construct a network. Whilst default behaviour for this is available, the full output of the testing is also provided, so that users may further explore the statistics using fdrtool. Functionality for visualising the covariance matrix is provided, as well as for the reconstructed network using igraph (Csardi and Nepusz, 2006).

Author(s)

Paul DW Kirk and Harry Gray

Maintainers: <[e-mail]> <[e-mail]>

References

Oba, S., Sato, M.A., Takemasa, I., Monden, M., Matsubara, K.I. and Ishii, S., 2003. doi.

Stacklies, W., Redestig, H., Scholz, M., Walther, D. and Selbig, J., 2007. doi.

Ilin, A. and Raiko, T., 2010. link

Porta, J.M., Verbeek, J.J. and Kroese, B.J., 2005. link

Strimmer, K., 2008. link.

Strimmer, K., 2008. doi.

Csardi, G. and Nepusz, T., 2006. link.

See Also

pcaMethods

HGray384/pcaNet documentation built on Nov. 14, 2020, 11:11 a.m.