# grep the name list for SVM training: protein, cell, interaction, process
# 1. "grep" against >=2 gram terms
# 2. the regular expressions are in Perl format
# 3. In most cases, the "grep" is omitted; in a few cases, when pipe is used, the complete shell command is listed, e.g.
#     grep -P 'cell' | grep -P 'lines?$'
# 4. Option -v stands for "NOT"

#Protein/Gene Names:
'(channels?)$'
'(receptors?)$'
'(transporters?)$'
grep -P '((enzymes?)|(ases?))( (.|I{1,3}|(IV)))?(-\d)?$' | grep -vP '(coenzyme)|((\t| )(base|phase)s?( |$))|(release$)'
'^protein|proteins?$' 
'(mutant )|(mutants?$)'
'complex(es)?$'
'CaM|hsp|(14-3-3|p53|p38|AChR|DnaK|EGFR|eNOS|eIF4H|ERK|GIRK|GRK|JNK|MAPK|SAPK|STAT|TFIIF)\d?$'
'G(i|o|s) ?(alpha|beta|gamma)'

#Cell Names:
'cells?$'
grep -P 'cell' | grep -P 'lines?$'
'((neuron)|(blast)|(cyte))s?$'

Interaction Names:
'binding$'
'phosphorylation$'
'activations?$'
'inhibitions?$'
'(interaction)|(merization)|(recognition)|( action)s?$'
'hydrolysis$'

#Biological Process Names:
'(activit(y|(ies)))$'
'release$'
'transfer$'
'exchange$'
'expressions?$'
'growth$'
grep -P '((repair)|(damage)|(apoptosis)|(sion)|(tion)|(synthesis)|(influx)|(efflux)|(import)|(shock)|(transit)|(transport)|(uptake))$' | grep -vP ' cation$'
grep -P '^((ca2)|(calcium))' | grep -vP '(cyt)|(calmodulin)|(CaM)'