The page you link to is a GenBank Flat file, the default format used by GenBank. So, to get that format, which includes the join
lines, use:
./efetch -db nuccore -id CP003820.1 -format gb
The join()
lines give the different ways in which a sequence can be built from this gene. For example, one mRNA sequence produced by the locus you mention (CP003820) is:
mRNA complement(join(100..1497,1556..1767,1823..2274,
2322..2782,2846..3126,3206..3890,3958..5263,5322..5422,
5494..>5645))
The corresponding coding sequence is for the mRNA above is:
CDS complement(join(168..1497,1556..1767,1823..2274,
2322..2782,2846..3126,3206..3890,3958..5263,5322..5422,
5494..5645))
These aren't necessarily splicing "modifications", and aren't "variants", they are simply the way that the exons are joined to make an mRNA.
Note that since CP003820.1 is "Cryptococcus neoformans var. grubii H99 chromosome 1, complete sequence", there are multiple genes described in there and each will have its own (or multiple, if there is alternative splicing) join
statements. Specifically, I see 1014 genes described in the file:
$ grep -Pc '^\s+gene\s+' CP003820.gb
1014
950 mRNAs:
$ grep -c 'mRNA.*join(' CP003820.gb
950
And 955 proteins:
$ grep -Pc 'protein_id=' CP003820.gb
955