Using gnu sed you can use a negated bracket expression:
gnu sed
sed 's/_[^\t]*//g' file
_[^\t]* matches underscore followed by zero or more non-tab characters.
_[^\t]*
try in awk:
awk '{gsub(/_[^\\]*/,"");print}' Input_file
try in sed:
sed 's/_[^\\]*//g' Input_file
EDIT: As OP said \t are the TABs in Input_file so adding following solution too now.
awk '{gsub(/_[^\t]*/,"");print}' Input_file
sed 's/_[^\t]*\t/\t/g'
if you're not worried about the last field.
sed 's/_[^\t]*\t/\t/g;s/_[^\t]*$//'
otherwise.