I have a bunch of transcript relative coordinates that I need to map to genomic coordinates. I am using a custom annotation file that I load using:

txdb <- makeTxDbFromGFF(file='../2022_10_04_orthoalltime_novel_dge/ortho_alltime_merged_temp.gtf',
                        dataSource="orthoalltime",
                        organism="Homo sapiens",
                        format='gtf')

I get all transcripts and exons in my annotation file:

tx_by_gene <- transcriptsBy(txdb, by="gene")
ex_by_tx <- exonsBy(txdb, "tx", use.names=TRUE)

I then get all exons belonging to my transcript of interest:

my_txid = 'MSTRG.26927.8'
my_transcripts <- tx_by_gene[[novel.small$my_gene_id[i]]]
my_tx_names <- mcols(my_transcripts)$tx_name
my_ex_by_tx <- ex_by_tx[intersect(my_tx_names, my_txid]

This is the output of my_ex_by_tx which is also the exon-level structure of 'MSTRG.26927.8'

GRangesList object of length 1:
$`19`
GRanges object with 18 ranges and 3 metadata columns:
       seqnames          ranges strand |   exon_id   exon_name exon_rank
          <Rle>       <IRanges>  <Rle> | <integer> <character> <integer>
   [1]        7 6109801-6109843      + |    228914        <NA>         1
   [2]        7 6111125-6111374      + |    228915        <NA>         2
   [3]        7 6115323-6115523      + |    228917        <NA>         3
   [4]        7 6135841-6135951      + |    228921        <NA>         4
   [5]        7 6139092-6139194      + |    228922        <NA>         5
   ...      ...             ...    ... .       ...         ...       ...
  [14]        7 6150412-6150506      + |    228932        <NA>        14
  [15]        7 6153756-6155195      + |    228933        <NA>        15
  [16]        7 6156754-6157055      + |    228934        <NA>        16
  [17]        7 6159450-6159493      + |    228936        <NA>        17
  [18]        7 6160555-6161564      + |    228937        <NA>        18
  -------
  seqinfo: 52 sequences (1 circular) from an unspecified genome; no seqlengths

If I try to map transcript/exon coordinates to genome coordinates:

coord <- GRanges(chromosome, IRanges(start = 1, end=43, names=my_txid), strand='+')
my_grange = mapFromTranscripts(coord, my_ex_by_tx)
my_grange

This successfully returns the first exon:

GRanges object with 1 range and 2 metadata columns:
                seqnames          ranges strand |     xHits transcriptsHits
                   <Rle>       <IRanges>  <Rle> | <integer>       <integer>
  MSTRG.26927.8        7 6109801-6109843      + |         1               1
  -------
  seqinfo: 52 sequences from an unspecified genome; no seqlengths

However, if I try to map position 44 onwards, the result does not include the second genomic region. Instead, it outputs the start of exon1 until the start of exon2 inclusive o the intron region:

coord <- GRanges(chromosome, IRanges(start = 1, end=44, names=my_txid), strand='+')
my_grange = mapFromTranscripts(coord, my_ex_by_tx)
my_grange
GRanges object with 1 range and 2 metadata columns:
                seqnames          ranges strand |     xHits transcriptsHits
                   <Rle>       <IRanges>  <Rle> | <integer>       <integer>
  MSTRG.26927.8        7 6109801-6111125      + |         1               1
  -------
  seqinfo: 52 sequences from an unspecified genome; no seqlengths

How do I get the genomic coordinates such that they are splice aware? The transcriptToGenome function from ensembldb returns the desired output but I cannot use it as my annotation database is not an ensdb object.