choose_k-method: Choice of the final classification of peaks
In FunChIP: Clustering and Alignment of ChIP-Seq peaks based on their shapes

Description Usage Arguments Details Value Author(s) See Also Examples

Selection of the final classification of the peaks, given the desired number of clusters and the presence or absence of the alignment procedure chosen. This choice is usually driven by the graph returned by cluster_peak.

1 2	## S4 method for signature 'GRanges' choose_k(object, k = NULL, shift.peak = NULL, cleaning = TRUE)

`object`	GRanges object. It must contain the metadata columns associated to the chosen classification. Further details are provided in `shift.peak`.
`k`	integer. Number of chosen clusters.
`shift.peak`	logical. If `TRUE`, the clustering with alignment is chosen, if `FALSE`, the classification without alignment is selected. If `shift.peak = TRUE`, `object` must contain the metadata columns `labels_shift`, `coef_shift`, `dist_shift`; if `shift.peak = FALSE` it must contain the columns `labels_NOshift` and `dist_NOshift`.
`cleaning`	logical. If `TRUE`, all metadata columns generated by FunChIP on the GRanges object are removed, and one new column containing the classification result is added: `cluster`. If `FALSE`, the metadata columns generated by FunChIP are kept. Default is `TRUE`. See Value for further details on the metadata column added.

The choice of the optimal number of clusters and the presence of alignment can be guided by the graph plotted in the cluster_peak method. In particular, for the optimal number of clusters k the distance significantly decreases with respect to the lower values of k, and negligibly increases with respect to higher values of k (elbow in the line). The introduction of the alignment leads to better clustering of the data if if the global distance is significantly lowered. registration.

if cleaning = FALSE, the GRanges object with a new metadata column:

cluster integer. The label correspondent to the classification chosen

If cleaning = TRUE, all the metadata columns added through all the analysis are removed and object is returned with just the metadata column cluster

Alice Parodi, Marco J. Morelli, Laura M. Sangalli, Piercesare Secchi, Simone Vantini

cluster_peak

# load the data
data(peaks)

# k = 3 clusters with the alignment
# with integer shifts are chosen

peaks.classified <- choose_k(peaks.data.cluster, k = 3, 
shift.peak = TRUE, cleaning = FALSE)