InferLandmark: Infer co-expression landmarks from cells' SR values
In ChenWeiyan/LandSCENT: Landscape Single Cell Entropy

Description Usage Arguments Value References Examples

This function identifies potency-coexpression clusters of single cells, called landmarks, and finally infers the dependencies of these landmarks which can aid in recontructing lineage trajectories in time course or development associated scRNA-Seq experiments.

InferLandmark(
  Integration.l,
  pheno.v = NULL,
  pctG = 0.01,
  Component_use = NULL,
  reduceMethod = c("PCA", "tSNE"),
  clusterMethod = c("PAM", "dbscan"),
  k_pam = 9,
  eps_dbscan = 10,
  minPts_dbscan = 5,
  pctLM = 0.05,
  pcorTH = 0.1
)

`Integration.l`	A list object from `InferPotency` function.
`pheno.v`	A phenotype vector for the single cells, of same length and order as the columns of `Integration.l$expMC`. Function can also automatically extract phenotype information from your original sce/cds data, please store the phenotype information under the name of `phenoInfo`.
`pctG`	A numeric. Percentage of all genes in `Integration.l$expMC` to select from each principal component in an SVD/PCA of `Integration.l$expMC`. The union set of all selected genes is then used for clustering. Default value is 0.01.
`Component_use`	A numeric. Specify the number of principal components in the PCA to use in the downstream analysis. Default value is NULL.
`reduceMethod`	A character, either "PCA" or "tSNE". Indicates using PCA or tSNE method to do dimension reduction.
`clusterMethod`	A character, either "PAM" or "dbscan". Indicates using dbscan or PAM method to do clustering.
`k_pam`	Only used when `clusterMethod` is set to be "PAM". Maximum number of co-expression clusters, when performing clustering. Default value is 9.
`eps_dbscan`	Only used when `clusterMethod` is set to be "dbscan". Size of the epsilon neighborhood. Default is 10.
`minPts_dbscan`	Only used when `clusterMethod` is set to be "dbscan". Number of minimum points in the eps region (for core points). Default is 5 points.
`pctLM`	Percentage of total number of single cells to allow as a minimum size for selecting interesting landmarks i.e. potency-coexpression clusters of single cells. Default value is 0.05.
`pcorTH`	Threshold for calling significant partial correlations. Default value is 0.1. Usually, single-cell experiments profile large number of cells, so 0.1 is a sensible threshold.

Integration.l A list incorporates the input list with a new list named InferLandmark.l.

InferLandmark.l A list contains twelve objects:

cl The co-expression clustering index for each single cell

pscl The potency coexpression clustering label for each single cell

distPSCL The distribution of single cell numbers per potency state and coexpression cluster

medLM A matrix of medoids of gene expression for the selected landmarks

srPSCL The average signaling entropy of single cells in each potency coexpression cluster

srLM The average signaling entropy of single cells in each landmark

distPHLM Table giving the distribution of single cell numbers per phenotype and landmark

cellLM Nearest landmark for each single cell

cellLM2 A vector specifying the nearest and next-nearest landmark for each single cell

adj Weighted adjacency matrix between landmarks with entries giving the number of single cells mapping closest to the two landmarks

pcorLM Partial correlation matrix of landmarks as estimated from the expression medoids

netLM Adjacency matrix of landmarks specifying which partial correlations are significant

selectGene Selected a group of genes for internal clustering

Teschendorff AE, Tariq Enver. Single-cell entropy for accurate estimation of differentiation potency from a cell’s transcriptome. Nature communications 8 (2017): 15599. doi: 10.1038/ncomms15599.

Teschendorff AE, Banerji CR, Severini S, Kuehn R, Sollich P. Increased signaling entropy in cancer requires the scale-free property of protein interaction networks. Scientific reports 5 (2015): 9646. doi: 10.1038/srep09646.

Banerji, Christopher RS, et al. Intra-tumour signalling entropy determines clinical outcome in breast and lung cancer. PLoS computational biology 11.3 (2015): e1004115. doi: 10.1371/journal.pcbi.1004115.

Teschendorff, Andrew E., Peter Sollich, and Reimer Kuehn. Signalling entropy: A novel network-theoretical framework for systems analysis and interpretation of functional omic data. Methods 67.3 (2014): 282-293. doi: 10.1016/j.ymeth.2014.03.013.

Banerji, Christopher RS, et al. Cellular network entropy as the energy potential in Waddington's differentiation landscape. Scientific reports 3 (2013): 3039. doi: 10.1038/srep03039.

data(Example.m)
data(net13Jun12.m)
Integration.l <- DoIntegPPI(exp.m = Example.m[, c(1:58,61:84,86:98,100)], ppiA.m = net13Jun12.m)
data(SR.v)
Integration.l$SR <- SR.v
InferPotency.o <- InferPotency(Integration.l)
InferLandmark.o <- InferLandmark(InferPotency.o)