Multi-line efetch digging

Question

I'm using efetch to dig out a middle "envelope" protein within a viral genome specifically within fasta format. The code below works but may not be robust for an entire database. Thus I want to replace the two "perl" lines with formal code within the Entrez package. I tried using efilter -query but couldn't get it to work. Genome positions are one approach but are not absolute from one virus to another. Rewriting this may be the best option.

ncbi2=NC_002021
efetch -db protein -format gpc -id "$ncbi2" |\
xtract -pattern INSDSeq -element INSDQualifier_value |\
perl -p -e 's/^(.+?)\s+taxon.*.*\s+\b([A-Z]{300}[A-Z]+).*/>$1\n$2/'|\
perl -p -e "s/[\t \;\/]+/_/g"

Output

>InfluenzaAvirus_A_PuertoRico_8_1934
MDVNPTLLFLKVPAQNAISTTFPYTGD.....

I will provide a more detailed explanation shortly. Any ideas welcome.

Answer 1

Not exactly what you were looking for but may be you can use something like this?

ncbi2=NC_002021
efetch -db protein -format gpc -id "$ncbi2" \
    | xtract -pattern INSDSeq_feature-table \
        -group INSDFeature \
        -if INSDFeature_key -equals 'mat_peptide' \
        -tab '\n' \
        -element INSDFeature_location,INSDQualifier_value \
    | grep 'Segment 2'

For explorations of the INSDSet XMLs, xtract has -insd option that you can use. In this particular case, you can do something like this:

efetch -db nuccore -id "$ncbi2" -format gpc \
    | xtract -insd mat_peptide product peptide

For more information, checkout EDirect documentation and some example queries