Thanks for the help. I've checked which data can be retrieved from the Homo.sapiens object using columns(Homo.sapiens), and there is actually information on the transcript start and transcript end. I think I will need to get the width first, and then modify the Homo.sapiens object somehow?

I checked that, and a TxDb object does not contain gene names... I would like to show the name of the genes not the transcripts names.

That definitely works but there must be an easier (in-built way, I hope). Anyway, getting the longest transcript is simple:

library(ggbio)
library(Homo.sapiens)

data(genesymbol, package = "biovizBase")

GetLongest <- function(Ranges){

  ## list transcripts
  tx <- transcripts(Homo.sapiens)

  ## list genes
  gns <- genes(Homo.sapiens)

  ## get transcripts overlapping the ranges
  subs <- subsetByOverlaps(tx, Ranges)

  ## overlap with genes to get gene names
  olap <- findOverlaps(subs, gns)

  ## some transcripts might not have a matched gene in the database
  subs <- subs[olap@from]

  ## add gene ID to the metadata:
  elementMetadata(subs) <- data.frame(elementMetadata(subs), 
                                      GENEID = as.character(gns[olap@to]$GENEID))

  ## sort by length and take the unique genes.
  ## that gives you the longest transcript per annotated gene
  subs <- subs[order(width(subs), decreasing = TRUE)]
  longest <- subs[!duplicated(subs$GENEID)]
  return(longest)
}

GetLongest(GRanges("chr10"  , IRanges(start = 78000000 , end = 79000000)))

GRanges object with 2 ranges and 3 metadata columns:
      seqnames            ranges strand |   TXID     TXNAME      GENEID
         <Rle>         <IRanges>  <Rle> | <list>     <list> <character>
  [1]    chr10 77542519-78317126      + |  38363 uc001jxi.3       83938
  [2]    chr10 78629359-79397577      - |  40147 uc001jxj.2        3778
  -------
  seqinfo: 93 sequences (1 circular) from hg19 genome