Something like this (this is untested):

while {} {

if (substr($_,0,1) eq '>') {

my @this = split / /,$_;

print "$this[0]\n";

} else {

print $_;

}

I am agree with Adam

Assuming that you are on unix/linux and don't have white spaces in the sequences themselves:

cut -d' ' -f1 < in.fasta

This may be much faster than the python and perl solutions.

TIMTOWTDI. As far as I understand you need the first and third fields only. You could do this with awk,

awk '{print $1" "$3}' file.fasta

or even with Perl,

while(){

if((my $v1, my $v2) = /^(\S+)\s+\S+\s+(\S+)/){

print "$v1 $v2\n";

}

perl -ne 'if (/^(>\S+)/){print "$1\n";}else{print $_;}' mysequences.fa

maybe is late, this script you can change the name of all sequences how you want. 

example: fastRename -in infield.fasta -o outfile.fasta -l new_label_

https://gist.github.com/cabraham03/a7d3457919b08ac556d8