Recent Releases of SEMgraph
SEMgraph - Copenhagen Interpretation
New development release of the CRAN package SEMgraph. News: * SEMrun() function. The algo ="cggm" based on high-dimensional GGGM is now implemented with the de-sparsified (de-biased) nodewise LASSO procedure applied on a Gaussian DAG model. The overall indices "deviance/df" and "srmr" are now computed using the observed correlation matrix also in p > n regime, where the estimated parameters are computed using the "regularized" (lambda corrected) correlation matrix.
SEMbap() function. New deconfounding methods to adjust the data matrix by removing latent sources of confounding encoded in them are implemented. The selected methods are either based on: (i) Bow-free Acyclic Paths (BAP) search (dalgo = "cggm" or "glpc"), (ii) LVs proxies as additional source nodes of the data matrix, Y (dalgo = "pc" or "glpc") or (iii) spectral transformation of Y (dalgo = "pc" or "trim").
SEMdag() function. New two-step DAG estimation from an input (or empty) graph, using in step 1) graph topological order or bottom-up search order, and in step 2) parent recovery with the LASSO-based algorithm are implemented. The estimate linear order are obtained from a priori graph topological vertex (LO = "TO") or level (LO = "TL") ordering, or with a data-driven vertex or level Bottom-up (LO = "BU") based on "glasso" residual variance ordering. The Top-Down (LO = "TD") is removed, being the BU more efficient to implement the topological search order.
Shipley.test() function. Added new argument cmax = Inf (default). This parameter can be used to perform only those tests where the number of conditioning variables does not exceed the given value. Output of the data.frame "dsep" has the same format of the localCI.test() function.
Various fixed bugs
- R
Published by fernandoPalluzzi about 3 years ago
SEMgraph - Quantum Tunneling
New development release of the CRAN package SEMgraph. News: * SEMbap() function changes. Now the function implements new deconfounding methods to adjust the data matrix by removing latent sources of confounding encoded in them. The selected methods are based on: (i) Bow-free Acyclic Paths (BAP) search, (ii) LVs proxies as additional source nodes of the data matrix Y or (iii) spectral transformation of Y.
Output dsep data.frame in Shipley.test() in SEMbap() functions has the same format of the localCI.test() function.
New model.learn() and extractClusters() functions
Correlation to graph utility
Mediation analysis utility
Various fixed bugs
- R
Published by fernandoPalluzzi over 3 years ago
SEMgraph - Gravitational Waves
New stable release of the CRAN package SEMgraph. News: - Added new parameterEstimates() function for parameter estimates output of a fitted SEM for RICF and CGGM algorithms similar to lavaan. - Updating summary.RICF() and summary.GGM() functions with parameterEstimates() - Updating hidden functions SEMricf(), SEMricf2(), SEMggm(), and SEMggm2() - Added in SEMrun() function the bootstrap resampling of SE (95% CI), and new argoment n_rep = 1000 (default) to set the bootstrap samples or permutation flip, if algo = "ricf", and the argumet SE = c("standard", "none"), if algo = "lavaan". - Fixed bugs in cplot, colorGraph, and SEMtree.
- R
Published by fernandoPalluzzi almost 4 years ago
SEMgraph - Gravitational Singularities
New stable release of the CRAN package SEMgraph. News: - Added new arguments for SEMdag() function: LO = "TO" or "TD" for knowledge-based topological order or data-driven top-down order, and penalty = TRUE or FALSE, separate penalty factors can be applied to each coefficient - Deprecated extendGraph() in favor of new resizeGraph() function, that re-sized graph, removing edges or adding edges/nodes if they are present or absent in a given reference network - Change modelSerch(), interactive procedure is out, and now a three step procedure is implemented for search strategies with new SEMdag() and resizeGraph functions - Change SEMgsa() deleting D,A,E p-values with more performing activation and inhibition pvalues - Added argument MCX2 = TRUE or FALSE for Shipley.test() function, a Monte Carlo P-value of the combined C test - Added new SEMdci() function for differentially connected genes inference - Change properties(), now extracted components are order by component sizes - Change argument q = q-quantile with q = 1-top/vcount(graph) in activeModule() function, now the induced graph for the "rwr" and "hdi" algorithms is defined by the top-n ranking nodes. - Various fixed bugs
- R
Published by fernandoPalluzzi over 4 years ago
SEMgraph - Quantum Fields
New stable release of the CRAN package SEMgraph. News: - Restored functions due to GGMncv upgrade from 2.0.0 to 2.1.0 - Bug Fixes
- R
Published by fernandoPalluzzi over 4 years ago
SEMgraph - Pillars of Creation
This is the first stable release of SEMgraph, also available on CRAN.
- R
Published by fernandoPalluzzi about 5 years ago
SEMgraph - Parallel Universes
This version includes more accurate and faster functions for model learning and large interactome weighting. Bow-free interaction search has been improved to enable a finer model adjustment.
- R
Published by fernandoPalluzzi over 5 years ago
SEMgraph - A New Start
This is the first release of the R package SEMgraph. It is endowed with the latest findings in causal inference, learning, and network-based discovery, in a fast, accurate, and easy-to-use framework. It is meant for computational biologists as well as anyone willing to experience a smooth yet rigorous adventure in network analysis and causality. SEMgraph is beta software: any suggestion and/or correction is welcome!
- R
Published by fernandoPalluzzi over 5 years ago