fix plot_cnv for contigs factors with NA labels

[jan-glx]

If plot_cnv is called with an infercnv object of which the C_CHR column of the gene_order slot is a factor which has NA as a label it fails like this:

STEP 15: Clustering samples (not defining tumor subclusters)

INFO [2020-07-17 23:02:32] define_signif_tumor_subclusters(p_val=0.1
INFO [2020-07-17 23:02:32] define_signif_tumor_subclusters(), tumor: 3
INFO [2020-07-17 23:02:36] cut tree into: 1 groups
INFO [2020-07-17 23:02:36] -processing 3,3_s1
INFO [2020-07-17 23:02:36] define_signif_tumor_subclusters(), tumor: 1
INFO [2020-07-17 23:04:08] cut tree into: 1 groups
INFO [2020-07-17 23:04:08] -processing 1,1_s1
INFO [2020-07-17 23:04:08] define_signif_tumor_subclusters(), tumor: 2
INFO [2020-07-17 23:04:33] cut tree into: 1 groups
INFO [2020-07-17 23:04:33] -processing 2,2_s1
INFO [2020-07-17 23:04:33] define_signif_tumor_subclusters(), tumor: 0
INFO [2020-07-18 04:04:39] cut tree into: 1 groups
INFO [2020-07-18 04:04:39] -processing 0,0_s1
INFO [2020-07-18 04:04:39] -mirroring for hspike
INFO [2020-07-18 04:04:39] define_signif_tumor_subclusters(p_val=0.1
INFO [2020-07-18 04:04:39] define_signif_tumor_subclusters(), tumor: spike_tumor_cell_0
INFO [2020-07-18 04:04:43] cut tree into: 1 groups
INFO [2020-07-18 04:04:43] -processing spike_tumor_cell_0,spike_tumor_cell_0_s1
INFO [2020-07-18 04:04:43] define_signif_tumor_subclusters(), tumor: simnorm_cell_0
INFO [2020-07-18 04:04:44] cut tree into: 1 groups
INFO [2020-07-18 04:04:44] -processing simnorm_cell_0,simnorm_cell_0_s1
INFO [2020-07-18 04:08:36] ::plot_cnv:Start
INFO [2020-07-18 04:08:36] ::plot_cnv:Current data dimensions (r,c)=10121,10773 Total=109784435.602225 Min=0.108562602149896 Max=9.87176157420995.
INFO [2020-07-18 04:08:38] ::plot_cnv:Depending on the size of the matrix this may take a moment.
INFO [2020-07-18 04:11:54] plot_cnv(): auto thresholding at: (0.506649 , 1.507125)

Error in rep(contig_name, contig_tbl[contig_name]) : invalid 'times' argument
5.
	
FUN(X[[i]], ...)
4.
	
lapply(contig_labels, function(contig_name) { rep(contig_name, contig_tbl[contig_name]) })
3.
	
unlist(lapply(contig_labels, function(contig_name) { rep(contig_name, contig_tbl[contig_name]) }))
2.
	
plot_cnv(infercnv_obj, k_obs_groups = k_obs_groups, cluster_by_groups = cluster_by_groups, cluster_references = cluster_references, out_dir = out_dir, title = "Preliminary infercnv (pre-noise filtering)", output_filename = "infercnv.preliminary", output_format = output_format, write_expr_matrix = TRUE, ...
1.
	
infercnv::run(infercnv_obj, cutoff = 0.1, out_dir = tempfile(), cluster_by_groups = TRUE, denoise = TRUE, HMM = TRUE)

While this is likely a rare edge case it can perhaps easily be fixed ( if this is the only issue) by simplifying the code like this.
To see how this works look at the following example code:

Normally everything is ok:

contig_tbl <- table(factor(c("1", "2", "3", "2", "3", "3", NA, NA) ))
contig_labels <- names(contig_tbl)
rep(names(contig_tbl), contig_tbl) # new
# [1] "1" "2" "2" "3" "3" "3"
unlist(lapply(contig_labels, function(contig_name) {rep(contig_name, contig_tbl[contig_name])})) # old
# [1] "1" "2" "2" "3" "3" "3"

But if some user thinks NA is an appropriate contig (chromosome) name the old code breaks:

contig_tbl <- table(factor(c("1", "2", "3", "2", "3", "3", NA, NA), exclude="" )) # allow NA labels
contig_labels <- names(contig_tbl)
rep(names(contig_tbl), contig_tbl) # new
# [1] "1" "2" "2" "3" "3" "3" NA  NA 
unlist(lapply(contig_labels, function(contig_name) {rep(contig_name, contig_tbl[contig_name])})) # old
#Error in rep(contig_name, contig_tbl[contig_name]) : 
#  invalid 'times' argument

I did not tests this patch locally & hope there is CI running that will check this.

kind regards,
Jan