1) Get the data
- Go to BioMart
- Select Ensembl Genes as database, Homo sapiens as dataset
- Then on the left panel click on Attributes then click on GENE
then thick gene start, gene end, transcript start, transcript end,
- on the left panel click on Filter then click on GENE then type your favorite gene ID in the appropriate field
- on top left panel click on Results then download.
2) Open it in Excel, R or whatever you like and substract the gene START and END to the transcripts START and END to get the position of transcripts relatively to the full gene. Done.
PS : Note that the START position is only the real start (like transcription start site) if the strand is +. When its on the minus strand, then the END position is the TSS.