searchInc: Search for spatial inconsistencies

Description

A function to test for spatial inconsistencies in a calibrated Hi-C distance matrix.

Usage

searchInc(calibHiC, hicCoord)

Arguments

calibHiC

A M*M Hi-C matrix,providing pairwise calibrated distances between M genomic bins. Zero off-diagonal values represent discarded/missing information. This matrix could be prepared with the function calibrate.

hicCoord

The genomic coordinates of the M Hi-C bins. A data frame with the following mandatory columns: chr (chromosome name), start (start position), end (end position). Additional columns are permitted but will not be used.

Value

neighborhoods: A list of length M, where neighborhoods[i] gives the neighborhood of i, if a spatial inconsistency was detected for i, and NULL otherwise. Here, the neighborhood of i is the sub-graph of immediate neighbors in trans, an igraph object with attributes to label nodes by their name (index in hicCoord), fullName (chromosome and genomic coordinates), membership (the connected component to which they belong, where i is always assigned with 0) and chr (chromosome name).
indices: The indices of loci (bins) with a spatial inconsistency.

Details

Denote delta(i,j) the calibrated Hi-C distances, then delta(i,j) define a weighted undirected graph G{V, E}, where V is the set of loci (bins) and E is the set of edges: {(vi,vj)| delta(i,j) > 0, i != j} with weights delta(i,j). We distinguish here between immediate neighbors from the same chromosome (cis) and from different chromosomes (trans) and detect a spatial inconsistency for a node v in G, if the sub-graph G' of all (immediate) trans neighbors of v is not connected. Further identifying the connected components in G' can highlight the cause of inconsistency and the underlying spatial division.

References

Y. Shavit, F.K. Hamey, P. Lio', FisHiCal: an R package for iterative FISH-based calibration of Hi-C data, 2014 (submitted).