Hi folks,

I am trying to plot.a zoomed in custom karyoploteR plot. The key issue with the zoomed in plots are that I need the loci a bit bigger which causes the gene labels to merge with each other.

To avoid this issue, I wished to plot only a few custom genes filtered based on the gene name. The key question is how would I do that in such a way that Txdb object makes its own filtered Txdb object which then I can feed it into kpPlotGenes. FYI, kpPlotGenes takes in Txdb object and author Bernat mentions that one can use custom Txdb directly for kpPlotGenes.

## TODO: Initialize the karyoplot
kp <- plotKaryotype(plot.type = 2, zoom = SNAPC3.region, cex = 1, plot.params = pp)
kpDataBackground(kp, data.panel = 1, col = "grey98")

## Plot genes
kpPlotGenes(kp, TxDb.Hsapiens.UCSC.hg19.knownGene, r0 = 0.2, r1 = 0.3, gene.name.cex = 0.8, data.panel = 2, gene.margin = 0, col = "darkmagenta", gene.name.col = "black", gene.name.position = "top", avoid.overlapping = TRUE)

The following ways I have tried and failed

1. Subset the gene.data object that I get from karyoploteR

genes.data <- makeGenesDataFromTxDb(TxDb.Hsapiens.UCSC.hg19.knownGene,
                                    karyoplot=kp,
                                    plot.transcripts = FALSE,
                                    plot.transcripts.structure = FALSE)
gene.data <- addGeneNames(genes.data)

selected.gene.data <- gene.data$genes[gene.data$genes$gene_id %in% ranges.df[ranges.df$coverage == unique(ranges.df$coverage)[1],]$names]

1. select() won't work here, but can I make it a custom Txdb object using something (this didn't work cause I had different column names)

txdb <- TxDb.Hsapiens.UCSC.hg19.knownGene
customtxdb <- select(txdb, keys = ranges.df[ranges.df$coverage == unique(ranges.df$coverage)[1],]$names, keytype="GENEID", columns = columns(txdb))
makeTxdb(customtxdb)

Error in check_colnames(transcripts, .REQUIRED_COLS, .OPTIONAL_COLS, "transcripts"): 'transcripts' must have at least the following cols: tx_id, tx_chrom,
  tx_strand, tx_start, tx_end

I think there is a better method in the documentation of genomic features on how to handle this, but I can't seem to find it.