Why are there many start_codon for the same gene in hg38.refGene.gtf

Question

[ManuelDB]

As far as I know, I was expecting around 20 thousand start_codon in the refGene.gtf file.

According to my genetics knowledge:

1. We have appr. 20 thousand protein-coding genes
2. there is only one start codon per protein_coding
3. Only protein-coding genes have start codon

If this is correct, why

ncbiRefSeq = ".GTF_files/hg38.refGene.gtf"

ncbiRefSeq = read_gtf(ncbiRefSeq)

ncbiRefSeq[ ("start_codon" == ncbiRefSeq['feature'])]

[Pierre Lindenbaum]

LIFE.

[ManuelDB]

Ahhh! I know about splicing but I though start codon refers to the gene not the transcript. I have seen that there are even more 5UTR why?

exon           836541
CDS            655401
5UTR           121178
transcript      84787
3UTR            67901
start_codon     64026
stop_codon      64026

the total length of gene is not directly provided in this gtf file?

[Pierre Lindenbaum]

I though start codon refers to the gene not the transcript.

that's a wrong assertion.

[Pierre Lindenbaum]

I have seen that there are even more 5UTR why?

a UTR can span more than one exon....