# Last-edition: 20060711 by ODiP # # Months # $month= { JAN:1 FEB:2 MAR:3 APR:4 MAY:5 JUN:6 JUL:7 AUG:8 SEP:9 OCT:10 NOV:11 DEC:12 } $rules={ # # Entry # entry: ~ { $In:[file:text] $Out pre{$Skip:0} } ('ID' {$entryFip= $Fip $Wrt} toln {$App} ('ID' {$Not} toln {$App})+ )? ~ # # Fields # fields: ~ {$In:entry $Out} id pdbid site_type header title keyword molecule+ synonym? ec_number? synthetic? org_sci? org_common? taxonomy? cell_loc? strain? variant? organ? tissue? cell? organelle? secretion? gene? go_proc_term? go_func_term? go_cell_comp_term? go? uniprot? embl? pir? transfac? ensembl? interpro? pfam? smart? panther? prints? tigr? tigrfams? hssp? hamap? prodom? kegg? kegg_comp? kegg_drug? pubchem? ligand? ligand_formul? drug? mol_chains num_site_chains site_descr res_notaa? num_of_aa exposure discont site_chains /.*/ {$Wrt} ~ id: ~ {$Wrt:id} 'ID ' toln ~ pdbid: ~ {$Wrt:pdbid} 'PDBID ' toln ~ site_type: ~ {$Wrt:site_type} 'SITE_TYPE ' toln ~ header: ~ {$Wrt:header} 'HEADER ' toln ~ title: ~ {$Wrt:title} 'TITLE ' toln ~ keyword: ~ {$Wrt:keyword} 'KEYWORD ' toln ~ molecule: ~ {$Wrt:molecule} 'MOLECULE ' toln ~ synonym: ~ {$Wrt:synonym} 'SYNONYM ' toln ~ ec_number: ~ {$Wrt:ec_number} 'EC_NUMBER ' toln ~ synthetic: ~ {$Wrt:synthetic} 'SYNTHETIC ' toln ~ org_sci: ~ {$Wrt:org_sci} 'ORGANISM_SCIENTIFIC ' toln ~ org_common: ~ {$Wrt:org_common} 'ORGANISM_COMMON ' toln ~ taxonomy: ~ {$Wrt:taxonomy} 'TAXONOMY ' toln ~ cell_loc: ~ {$Wrt:cell_loc} 'CELLULAR_LOCATION ' toln ~ strain: ~ {$Wrt:strain} 'STRAIN ' toln ~ variant: ~ {$Wrt:variant} 'VARIANT ' toln ~ organ: ~ {$Wrt:organ} 'ORGAN ' toln ~ tissue: ~ {$Wrt:tissue} 'TISSUE ' toln ~ cell: ~ {$Wrt:cell} 'CELL ' toln ~ organelle: ~ {$Wrt:organelle} 'ORGANELLE ' toln ~ secretion: ~ {$Wrt:secretion} 'SECRETION ' toln ~ gene: ~ {$Wrt:gene} 'GENE ' toln ~ go_proc_term: ~ {$Wrt:go_proc_term} 'GO_PROCESS_TERM ' toln ~ go_func_term: ~ {$Wrt:go_func_term} 'GO_FUNCTION_TERM ' toln ~ go_cell_comp_term: ~ {$Wrt:go_cell_comp_term} 'GO_CELLULAR_COMPONENT_TERM ' toln ~ go: ~ {$Wrt:go} 'GO ' toln ~ uniprot: ~ {$Wrt:uniprot} 'UNIPROT ' toln ~ embl: ~ {$Wrt:embl} 'EMBL ' toln ~ pir: ~ {$Wrt:pir} 'PIR ' toln ~ transfac: ~ {$Wrt:transfac} 'TRANSFAC ' toln ~ ensembl: ~ {$Wrt:ensembl} 'ENSEMBL ' toln ~ interpro: ~ {$Wrt:interpro} 'INTERPRO ' toln ~ pfam: ~ {$Wrt:pfam} 'PFAM ' toln ~ smart: ~ {$Wrt:smart} 'SMART ' toln ~ panther: ~ {$Wrt:panther} 'PANTHER ' toln ~ prints: ~ {$Wrt:prints} 'PRINTS ' toln ~ tigr: ~ {$Wrt:tigr} 'TIGR ' toln ~ tigrfams: ~ {$Wrt:tigrfams} 'TIGRFAMS ' toln ~ hssp: ~ {$Wrt:hssp} 'HSSP ' toln ~ hamap: ~ {$Wrt:hamap} 'HAMAP ' toln ~ prodom: ~ {$Wrt:prodom} 'PRODOM ' toln ~ kegg: ~ {$Wrt:kegg} 'KEGG ' toln ~ kegg_comp: ~ {$Wrt:kegg_comp} 'KEGG_COMPOUND ' toln ~ kegg_drug: ~ {$Wrt:kegg_drug} 'KEGG_DRUG ' toln ~ pubchem: ~ {$Wrt:pubchem} 'PUBCHEM ' toln ~ ligand: ~ {$Wrt:ligand} 'LIGAND ' toln ~ ligand_formul: ~ {$Wrt:ligand_formul} 'LIGAND_FORMUL ' toln ~ drug: ~ {$Wrt:drug} 'DRUG ' toln ~ mol_chains: ~ {$Wrt:mol_chains} 'MOL_CHAINS ' toln ~ num_site_chains: ~ {$Wrt:num_site_chains} 'NUM_SITE_CHAINS ' toln ~ site_descr: ~ {$Wrt:site_descr} 'SITE_DESCR ' toln ~ res_notaa: ~ {$Wrt:res_notaa} 'RESIDUE_NOTAA ' toln ~ num_of_aa: ~ {$Wrt:lensite} 'NUMBER_OF_AA' spaces integer {$Out:lensite $Wrt} spaces integer {$Out:lensurr $Wrt} ln {$Out:fields} ~ exposure: ~ {$Wrt:expsite} 'EXPOSURE' spaces real {$Out:expsite $Wrt} spaces real {$Out:expsurr $Wrt} ln {$Out:fields} ~ discont: ~ {$Wrt:discont} 'DISCONTINUITY ' real {$Out:discont $Wrt} ln {$Out:fields} ~ site_chains: ~ {$Wrt:site_chains} 'SITE_CHAINS ' toln ~ # # Index # i_id: ~ {$In:[fields c:id] $Out:id} tag_sp fill {$Wrt} ln ~ i_pdbid: ~ {$In:[fields c:pdbid] $Out:pdbid} tag_sp word {$Wrt} ~ i_site_type: ~ {$In:[fields c:site_type] $Out:site_type} tag_sp word {$Wrt} ~ i_header: ~ {$In:[fields c:header] $Out:header} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_title: ~ {$In:[fields c:title] $Out:title} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_keyword: ~ {$In:[fields c:keyword] $Out:keyword} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_molecule: ~ {$In:[fields c:molecule] $Out:molecule} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_synonym: ~ {$In:[fields c:synonym] $Out:synonym} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_ec_number: ~ {$In:[fields c:ec_number] $Out:ec_number} tag_sp ( /[0-9.*-]+/ {$Wrt} | /[\/, ]+/ )+ ln ~ i_synthetic: ~ {$In:[fields c:synthetic] $Out:synthetic} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_org_sci: ~ {$In:[fields c:org_sci] $Out:org_sci} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_org_common: ~ {$In:[fields c:org_common] $Out:org_common} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_taxonomy: ~ {$In:[fields c:taxonomy] $Out:taxonomy} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_cell_loc: ~ {$In:[fields c:cell_loc] $Out:cell_loc} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_strain: ~ {$In:[fields c:strain] $Out:strain} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_variant: ~ {$In:[fields c:variant] $Out:variant} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_organ: ~ {$In:[fields c:organ] $Out:organ} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_tissue: ~ {$In:[fields c:tissue] $Out:tissue} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_cell: ~ {$In:[fields c:cell] $Out:cell} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_organelle: ~ {$In:[fields c:organelle] $Out:organelle} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_secretion: ~ {$In:[fields c:secretion] $Out:secretion} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_gene: ~ {$In:[fields c:gene] $Out:gene} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_go_proc_term: ~ {$In:[fields c:go_proc_term] $Out:go_proc_term} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_go_func_term: ~ {$In:[fields c:go_func_term] $Out:go_func_term} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_go_cell_comp_term: ~ {$In:[fields c:go_cell_comp_term] $Out:go_cell_comp_term} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_go: ~ {$In:[fields c:go] $Out:go} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_uniprot: ~ {$In:[fields c:uniprot] $Out:uniprot} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_embl: ~ {$In:[fields c:embl] $Out:embl} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_pir: ~ {$In:[fields c:pir] $Out:pir} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_transfac: ~ {$In:[fields c:transfac] $Out:transfac} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_ensembl: ~ {$In:[fields c:ensembl] $Out:ensembl} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_interpro: ~ {$In:[fields c:interpro] $Out:interpro} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_pfam: ~ {$In:[fields c:pfam] $Out:pfam} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_smart: ~ {$In:[fields c:smart] $Out:smart} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_panther: ~ {$In:[fields c:panther] $Out:panther} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_prints: ~ {$In:[fields c:prints] $Out:prints} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_tigr: ~ {$In:[fields c:tigr] $Out:tigr} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_tigrfams: ~ {$In:[fields c:tigrfams] $Out:tigrfams} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_hssp: ~ {$In:[fields c:hssp] $Out:hssp} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_hamap: ~ {$In:[fields c:hamap] $Out:hamap} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_prodom: ~ {$In:[fields c:prodom] $Out:prodom} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_kegg: ~ {$In:[fields c:kegg] $Out:kegg} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_kegg_comp: ~ {$In:[fields c:kegg_comp] $Out:kegg_comp} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_kegg_drug: ~ {$In:[fields c:kegg_drug] $Out:kegg_drug} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_pubchem: ~ {$In:[fields c:pubchem] $Out:pubchem} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_ligand: ~ {$In:[fields c:ligand] $Out:ligand} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_ligand_formul: ~ {$In:[fields c:ligand_formul] $Out:ligand_formul} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_drug: ~ {$In:[fields c:drug] $Out:drug} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_mol_chains: ~ {$In:[fields c:mol_chains] $Out:mol_chains} tag_sp ( /[^\n]/ {$Uniq} )+ ln ~ i_num_site_chains: ~ {$In:[fields c:num_site_chains] $Out:num_site_chains} tag_sp integer {$Wrt} ln ~ i_site_descr: ~ {$In:[fields c:site_descr] $Out:site_descr} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_res_notaa: ~ {$In:[fields c:res_notaa] $Out:res_notaa} tag_sp ( word { if:($StrLen:$Ct >= 3) {$Wrt} } | nonword )* ~ i_discont: ~ {$In:[fields c:discont] $Out:discont} tag spaces real {$Wrt} ln ~ i_site_chains: ~ {$In:[fields c:site_chains] $Out:site_chains} tag_sp ( /[^\n]/ {$Uniq} )+ ln ~ # # Html # h_pdbid: ~ {$In:[fields c:pdbid t:html] pre {if:$ParInt:isTable $Fail} } tag_sp word { $Rep:$Hlink:[entryR p:{($ParStr:userIdOpt) 'ENPDB' 'id' $Ct $Ct}] } ln ~ # # Other # ln: ~ '\n' ~ toln: ~ /[^\n]*\n/ ~ sp_toln: ~ / [^\n]*\n/ ~ spaces: ~ / +/ ~ tag: ~ /[A-Z0-9_-]+/ ~ tag_sp: ~ /[A-Z0-9_-]+ / ~ word: ~ /[a-zA-Z0-9]+/ ~ nonword: ~ /[^a-zA-Z0-9]*/ ~ wordext: ~ /[a-zA-Z0-9_-]+/ ~ fill: ~ /[^ \t\n]+/ ~ nonfill: ~ /[ \t\n]*/ ~ real: ~ '-'? /[0-9]+\.[0-9]+/ ~ integer: ~ '-'? /[0-9]+/ ~ htag: ~ /<[^\<>]+>/ ~ } # rulesSRS error