Extract text between symbols with bash - awk

Question

First off, I am relatively new at this, so please bear with me.
I have an annotated transcriptome .fasta file, which contains ~60,000 records of genes like these two:
>comp35897_c0_seq11 len=1039 path=[11:0-12;24:13-1038] Match_Acc=E5SX33 Gene=Putative_CAP-Gly_domain_protein
TTTTAAATTGATTACTTTGCTATTTTTGGCAATGTTGGACTGAGTTGTCGTATTTTTTCG
>comp32620_c0_seq3 len=1874 path=[1:0-195;197:196-220;222:221-354;356:355-481;4197:482-487;489:488-579;581:580-1159;1161:1160-1712;1714:1713-1729;1731:1730-1794;5873:1795-1873] Match_Acc=K1PQJ1 Gene=HAUS_augmin-like_complex_subunit_3 GO=GO:0051225,GO:0070652
CAGACTTTTGGATTTAGTACATGTATGTATGAATATGTGTTTCAATGTACAACTCAGGAT
I am trying to create a two-column, space-delimited .tab with component number in the first column and gene name in the second column. I have looked at many similar posts using grep, sed, or awk, but none of the suggested code has worked for me.
Specifically, what I need to pull from the .fasta is the comp number between the > and the next space for the first column, and gene name between Gene=and the next space. For the two genes above, that should give me:
comp35897_c0_seq11 Putative_CAP-Gly_domain_protein
comp32620_c0_seq3 HAUS_augmin-like_complex_subunit_3
Any help would be much appreciated!

Answer 1

with awk:
Skip's gene name if 'gene' is absent
awk 'BEGIN{RS=">"} NF>1{if($5 ~ /Gene=/){gsub("Gene=","",$5); print $1,$5} else {print $1}}' < transcriptome.fasta > space-delimited.tab
Output:
comp35897_c0_seq11 Putative_CAP-Gly_domain_protein
comp32620_c0_seq3
Skip's the record if 'gene' is absent
awk 'BEGIN{RS=">"} NF>1{if($5 ~ /Gene=/){gsub("Gene=","",$5); print $1,$5}}' < transcriptome.fasta > space-delimited.tab
Output:
comp35897_c0_seq11 Putative_CAP-Gly_domain_protein

Answer 2

Have you tried anything yet?
with sed you could do:
sed 's/>\(comp[^ ]\+\) \+.*Gene=\([^ ]\+\) .*$/\1 \2/'
which looks complex but is relatively easy to understand if you take it slowly and break it down into it's component parts.
edit
ok, so to ensure sed only outputs what you want, you need to switch on the 'no output by default' mode -n and explicitly print each line you are interested in p
I'll try to break it down, so that it is understandable.
comp[^ ]\+ #is a regex that says:
#text that starts with the string 'comp'
#and is followed by at least one character
#that is anything that isn't a space (the [^ ])
\(comp[^ ]\+\) #is the sed construct that remembers what
#that regex matches.
.* #is the regex for zero or more of any chars.
'Gene=\([^ ]\+\) ' #look for the string 'Gene' followed by an
#equals sign, followed by at least one char
#that isn't a space, followed by a space
#oh, and remember the bit after = and before the space
so, along with the -n and the p switches for sed you could use:
sed -n 's/>\(comp[^ ]\+\) \+.*Gene=\([^ ]\+\) .*$/\1 \2/p'