GenBank annotation

[latentbird]

Greetings!
I'm working with GenBank annotated files.
The feature key source has a mandatory identifier /mol_type
It can have a few values, among them there are 'genomic DNA' and 'tRNA'. I'm interested in genes, coding for tRNA. But in both cases, the origin (sequence in genBank record) is the same. For instance:

LOCUS NC_003076 72 bp DNA linear PLN 14-AUG-2008
DEFINITION Arabidopsis thaliana chromosome 5, complete sequence.
ACCESSION NC_003076 REGION: 26938898..26938969
...
FEATURES Location/Qualifiers
source 1..72
/organism="Arabidopsis thaliana"
/mol_type="genomic DNA"
/db_xref="taxon:3702"
/chromosome="5"
/ecotype="Columbia"
gene complement(<1..>72)
/locus_tag="AT5G67460"
/gene_synonym="K9I9.2; K9I9_2"
/db_xref="GeneID:836882"
/db_xref="TAIR:AT5G67460"
mRNA complement(<1..>72)
/locus_tag="AT5G67460"
/gene_synonym="K9I9.2; K9I9_2"
/note="supported by ESTs: gi|124736715|gb|EH828117|,
gi|5851437|gb|AW004408|, gi|124921280|gb|EH994912|"
/transcript_id="NM_126146.2"
/db_xref="GI:30698308"
/db_xref="GeneID:836882"
/db_xref="TAIR:AT5G67460"
gene 1..72
/locus_tag="AT5G67455"
/gene_synonym="67938.TRNA-MET-1"
/db_xref="GeneID:3771575"
/db_xref="TAIR:AT5G67455"
tRNA 1..72
/locus_tag="AT5G67455"
/gene_synonym="67938.TRNA-MET-1"
/product="tRNA-Met"
/note="pre-tRNA;
tRNA-Met (anticodon: CAT)"
/transcript_id="NR_023309.1"
/db_xref="GI:186532837"
/db_xref="GeneID:3771575"
/db_xref="TAIR:AT5G67455"
ORIGIN
1 atcagagtgg cgcagcggaa gcgtggtggg cccataaccc acaggtccca ggatcgaaac
61 ctggctctga ta
//

and

LOCUS NR_023309 72 bp tRNA linear PLN 22-MAY-2008
DEFINITION Arabidopsis thaliana (AT5G67455) tRNA.
ACCESSION NR_023309
...
FEATURES Location/Qualifiers
source 1..72
/organism="Arabidopsis thaliana"
/mol_type="tRNA"
/db_xref="taxon:3702"
/chromosome="5"
/ecotype="Columbia"
gene 1..72
/locus_tag="AT5G67455"
/gene_synonym="67938.TRNA-MET-1"
/db_xref="GeneID:3771575"
/db_xref="TAIR:AT5G67455"
ORIGIN
1 atcagagtgg cgcagcggaa gcgtggtggg cccataaccc acaggtccca ggatcgaaac
61 ctggctctga ta
//

Two records describe the same gene, have the same sequence, but /mol_type is different

So, the question is - can I use the second record (/mol_type = 'tRNA') for gene nucleotide content analysis?

[latentbird]

Just in case, if somebody encounters the same problem...
I've sent a question to NCBI and they answered the following:

These sequences are from the genomic DNA and are known to encode tRNA genes. There are so many tRNA gene sequences available with good sequence conservation that no sequencing center or lab (theoretically) has to try to isolate tRNA and reverse engineer it.