The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.
BIBLIOME / Lingua-BioYaTeA-0.11 / etc / bioyatea / post-processing-filtering.conf 
# que des chiffres et des blancs, tirets, = ...
#CLEAN=FORM::/^[0-9]*([- \.°,×\^=][0-9]*)?/
CLEAN=FORM::/^[0-9\- \.°,×\^=]*$/
#numbers and points
#CLEAN=FORM::/^[0-9]*(\.|,|°|-|×|\^|:|=)?([0-9]+)?(\.|,|°|-|×|\^)?([0-9]+)?$/
#starts with numbers
#CLEAN=FORM::/^\[*[0-9][0-9]*[ \(\[\°\^\+Xx]*[0-9]*[A-Z]*/
#starts with numbers and letters
#CLEAN=FORM::/^[0-9][0-9]*[A-Za-z][A-Za-z]*[0-9]* [^expression|peptide|cDNA|complex|mice|gene|mouse|protein|cell|antigen|polypeptide|polynucleotide|mRNA|acid|.*dimer|group|GST|variant|mutant|inhibitor|receptor|antibod|function|sequence|activity]/
#numbers and letters
#CLEAN=FORM::/^[0-9]+( \.|,|\=|°|×|\^:)?([A-Za-z]+)?$/
CLEAN=FORM::/^[0-9]+( \.|,|°|×|\+\^:)? ?[A-Za-z]*(-[a-z])?$/
#contains numbers
CLEAN=FORM::/[0-9][0-9]* [0-9][0-9]* [0-9][0-9]*/
#letters and points
CLEAN=FORM::/^[A-Za-z]+\.$/
#just  numbers
#CLEAN=FORM::/^[0-9]*$/
#just  one or two letters
CLEAN=FORM::/^[A-Za-z]{0,2}$/
#starts with letter and number
CLEAN=FORM::/^[A-Za-z]{0,2}$/
#starts with letter
CLEAN=FORM::/^[A-Za-z][0-9\.][0-9]* [A-Za-z][0-9\.]*$/
#starts with letter
CLEAN=FORM::/^[A-Za-z][0-9\.][0-9\.]*$/
#starts with letter
#CLEAN=FORM::/^[A-Za-z ][^A-Za-z .* ?peptide|region|gene|cell|protein]/
# sequence ADN
CLEAN=FORM::/^[ATCG \->53\']{8,}$/
CLEAN=FORM::/^[MGFSIPLVMKRQHWAXY0-9]{12,}$/
# date (01/01/2008)
CLEAN=FORM::/^\d+\/\d+\/\d+$/
# number between []
CLEAN=FORM::/^[\[\(]\d+[\]\)]$/
# number followed by a letter (with/without space)
#CLEAN=FORM::/^[0-9]+\.?([0-9]*)? ?[A-Za-z]+?$/
CLEAN=FORM::/^[0-9]+\.?[0-9]* ?[A-Za-z]*$/
# sth followed by g/ mL, mg/ mL, mg/ ml...
CLEAN=FORM::/^(g\/ mL|mg\/ mL|mg\/ Ml|mm|mM|µm|µl|ml|Pa |-ml|µM|M |mL|nmol|[^ ]nmol|mmol|N |liters|wt|x |°C|Å|RPM|pfu|mg|[0-9]+g| g)/
# starts with nombre followed by g/ mL, mg/ mL, mg/ ml...
#CLEAN=FORM::/^[0-9]+( |\.|=|:|-|\^-|\^|×|,)?[0-9]*(g\/ mL|mg\/ mL|mg\/ Ml|mm|mM|µm|µl|ml|-ml|Pa|µM|g|s|mg|M|mL|nmol|mmol|N|liters|wt|x|°C|Å|RPM|nM|L|µm|atoms\/ sec|cells\/ ml|cells\/ mL|liters\/ mole|min|HOBt\/ NMP|cm2|cm|sec|day|h|H|th|st|nd|MHz|P|week|fold|cells\/ well|hour|membered|-membered|labeled|-labeled|position|cells|°F|pfu|meter|time)/
CLEAN=FORM::/^[0-9][ \.=:-\^-\^×,0-9]*(g\/ mL|mg\/ mL|mg\/ Ml|mm|mM|µm|µl|ml|-ml|Pa|µM|g|s|mg|M|mL|nmol|mmol|N|liters|wt|x|k|kN|°C|Å|RPM|nM|L|µm|atoms\/ sec|cells\/ ml|cells\/ mL|liters\/ mole|min|HOBt\/ NMP|cm2|cm|sec|mdx|day|h|H|th|st|nd|MHz|P|week|fold|cells\/ well|hour|membered|-membered|labeled|-labeled|position|cells|°F|pfu|meter|time)$/
# journals
CLEAN=FORM::/J\. Biol/
# starts with + and a number
#CLEAN=FORM::/^\+ ?[0-9]/
# starts with +
CLEAN=FORM::/^\+/
# starts with -
CLEAN=FORM::/^-/
# starts with 0=
CLEAN=FORM::/^0=/
# starts with >, ...,* ",^,|,>,<%-
CLEAN=FORM::/^[%,\.â¤\#\*\&\*\""\/\^\^\|\<\<\>\>\>=\±\+\'\°\~=-]/
# starts with letters or/and numbers and )
#CLEAN=FORM::/^([A-Za-z]+|[0-9]+)(\.|,|°|-|×|\^)?([A-Za-z]+|[0-9]+)?(\.|,|°|-|×|\^)?([A-Za-z]+|[0-9]+)?\)/
CLEAN=FORM::/^[^\(]*\)/
# starts with letters or/and numbers and ]
#CLEAN=FORM::/^([A-Za-z]+|[0-9]+)(\.|,|°|-|×|\^)?([A-Za-z]+|[0-9]+)?(\.|,|°|-|×|\^)?([A-Za-z]+|[0-9]+)?\]/
CLEAN=FORM::/^[^\[]*\]/
# starts with letters or/and numbers and ^ and numbers/letters
#CLEAN=FORM::/^(\w+|\d+)\^(\w+|\d+)?$/
CLEAN=FORM::/^\w+\^\w*$/
# starts with letters and - and numbers (à réfléchir si on garde ou pas; on risque d'enlever des noms de protéines etc.)
#CLEAN=FORM::/^[A-Za-z]{0,2}-([0-9]+)?$/
# starts with letters and/without . and/without space and numbers with/without . (some entities can be relevant,ex. codon 9 ou gentamicine 5)
#CLEAN=FORM::/^[A-Za-z]+(\.| )[0-9]+\.?([0-9]+)?$/
CLEAN=FORM::/^[A-Za-z]+[\. ][0-9]+\.?[0-9]*$/
#ends by /
CLEAN=FORM::/\/$/
#ends by \
CLEAN=FORM::/\\$/
#ends by (
CLEAN=FORM::/\($/
#ends by .%^,-/
CLEAN=FORM::/[\.%,-\/\+[^spp?\.]]$/
#ends by (and letters or/and numbers
#CLEAN=FORM::/.*\(([A-Za-z]+|[0-9]+)(\.|,|°|-|×|\^)?([A-Za-z]+|[0-9]+)?(\.|,|°|-|×|\^)?([A-Za-z]+|[0-9]+)?$/
CLEAN=FORM::/\([^\)]*$/
#ends by [and letters or/and numbers
#CLEAN=FORM::/.*\[([A-Za-z]+|[0-9]+)(\.|,|°|-|×|\^)?([A-Za-z]+|[0-9]+)?(\.|,|°|-|×|\^)?([A-Za-z]+|[0-9]+)?$/
CLEAN=FORM::/\[[^\]]*$/
#ends by et al
CLEAN=FORM::/et al$/
#contains et al
CLEAN=FORM::/et al/
#et
CLEAN=FORM::/^et$/
#et
CLEAN=FORM::/ et$/
#et al
CLEAN=FORM::/^et al$/
#contains ,*,~,°/
#CLEAN=FORM::/^.*[\*\~\°\#½\/%\<\<\>\>\>=\<=].*$/
CLEAN=FORM::/[\*\~\°\\/\<\<\>\>]/
#CLEAN=FORM::ends by =
CLEAN=FORM::/=$/
#starts by a number and a letter or +,-
#CLEAN=FORM::/^\d+ ?(\w+)? ?[%\*=+-\/\)]? ?(\d+)?/
# measures
CLEAN=FORM::/^[0-9]+[\- ](bp|kDa|Da|fold|kb|kilobase|base-pair|old|aa|amino-acid|K|codon|dalton|base pair|times|kilodalton|min)$/
# translation
CLEAN=FORM::/-->/
# ->
CLEAN=FORM::/->/
# > < =
CLEAN=FORM::/^[=<>]/
# position on ADN
CLEAN=FORM::/positions? [\+\-]?[0-9]+/
# vs. 
CLEAN=FORM::/^vs/
# ends with (SEQ
CLEAN=FORM::/\(SEQ$/
# contains table, Table, TABLE
CLEAN=FORM::/^(table|Table|TABLE)/
# contains invention
CLEAN=FORM::/(invention|Invention|INVENTION)/
# contains level
#CLEAN=FORM::/^(level|levels |Level|Levels|LEVEL|LEVELS)\.?$/
#ends with and/
CLEAN=FORM::/(and|AND)\/$/
# contains FIG. number
CLEAN=FORM::/FIG\.([0-9]+)?/
# contains Preferred
#CLEAN=FORM::/(P|p)referred/
CLEAN=FORM::/[Pp]referred/
# contains Abstract
CLEAN=FORM::/Abstract/
# contains Image
CLEAN=FORM::/[Ii]mage[s]?/
# contains NO:
CLEAN=FORM::/NO\:/
# contains claim
#CLEAN=FORM::/(claim|Claim|claims|Claims)/
CLEAN=FORM::/[Cc]laim[s]?/
# contains size
#CLEAN=FORM::/(size|Size|SIZE)/
#delete all forms that the occurrence's number = 1
#CLEAN=NUMBER_OCCURRENCES::/^1$/
# contains method
CLEAN=FORM::/[Mm]ethod[s]?[^method of Schantz method of Towbin Rossler method]/
# contains variety of
#CLEAN=FORM::/[Vv]ariet[yies]?/
# contains first, second, third
#CLEAN=FORM::/(first|second|third)/
# contains approach(es)
#CLEAN=FORM::/[Aa]pproach[es]?/
# starts with wt,cm 
CLEAN=FORM::/^(wt|cm|[Kk]g|[Kk]hz|[Mm]g|m[Ll]|m[Mm]|mmole?s?|ng|n[Mm]|nmole?s?|pmole?s?|sec ) /
# contains point of view
CLEAN=FORM::/[Pp]oint of view/
# contains wide, useful
#CLEAN=FORM::/([Ww]ide|[Uu]seful|[Tt]ypical)/
# contains each, this, these...
CLEAN=FORM::/([Ee]ach |[Tt]his |[Tt]hese |[Tt]hose )/
# contains kind
#CLEAN=FORM::/[Kk]ind[s]?/
# contains use of
#CLEAN=FORM::/[^ca]use of/
# startswith number and (number)
CLEAN=FORM::/^[0-9]+\( ?[0-9]+\)/
# startswith number and (letter)
CLEAN=FORM::/^[0-9]+\( ?[A-Za-z]+\)/
# startswith number and (number and letter)
CLEAN=FORM::/^[0-9]+\([0-9]+[A-Za-z]\)/
# starts with <number
CLEAN=FORM::/^\<[0-9]+/
# starts with > number
CLEAN=FORM::/^\>[0-9]+/
# starts with [number]
CLEAN=FORM::/^\[[0-9]+\]/
# starts with [sth number]
CLEAN=FORM::/^\[.*[0-9]\]/
# contains '
CLEAN=FORM::/'/
# starts with aa ou AA
#CLEAN=FORM::/^(AA|aa)\.?/
# starts with (in)ability
#CLEAN=FORM::/^[in]?ability/ 
# contains type of
#CLEAN=FORM::/[Tt]ype[s]? of/
# contains various
CLEAN=FORM::/[Vv]arious/
# contains volume
#CLEAN=FORM::/volume/
# contains terms
CLEAN=FORM::/terms/
# contains whole sauf whole blood
#CLEAN=FORM::/[Ww]hole [^whole blood]/
# contains cfu
#CLEAN=FORM::/cfu/
# contains above
CLEAN=FORM::/above/
# contains absolute
#CLEAN=FORM::/[Aa]bsolute/
# contains abundant
#CLEAN=FORM::/[Aa]bundant/
# contains academic
#CLEAN=FORM::/[Aa]cademic/
# contains acatt
#CLEAN=FORM::/[Aa]catt/
# contains acceptable
#CLEAN=FORM::/[Aa]cceptable/
# contains accepted
#CLEAN=FORM::/[Aa]ccepted/
# contains accessible
#CLEAN=FORM::/[Aa]ccessible/
# contains accession, access
#CLEAN=FORM::/[Aa]ccess[ion]?/
# contains acoustic
#CLEAN=FORM::/[Aa]coustic/
# contains acquired, acquisition
#CLEAN=FORM::/[Aa]cqui(red|sition)/
# contains significant
#CLEAN=FORM::/[Ss]ignificant/
# contains example
CLEAN=FORM::/[Ee]xample[s]?/
# contains added,additional 
#CLEAN=FORM::/[Aa]dd(ed|itional)/
# contains adequate,adjacent,available,availability 
#CLEAN=FORM::/[Aa]d(equate|jacent|joining|junct|vanced|vantages?|vantageous)/
# contains aforementioned 
CLEAN=FORM::/[Aa]forementioned/
# contains  aggressive,authority
#CLEAN=FORM::/[Aa](ggressive|uthority)/
# starts with a ou A 
CLEAN=FORM::/^[Aa] /
# contains allowable,ambient,amountamplitude,attractive 
#CLEAN=FORM::/[Aa](llowable|mounts?)/
# contains analogous,applicable,apparent
#CLEAN=FORM::/[Aa](nalogous|pparent|pplicable|pplied|ppreciable|ppropriate)/
# contains art of, part of, starts of
#CLEAN=FORM::/[Aa]rt of[^heart]/
# contains bad
#CLEAN=FORM::/[Bb]ad /
# starts with BY
CLEAN=FORM::/^[Bb][Yy] /
# contains beginning,brief, better
CLEAN=FORM::/[Bb](etter)/
# contains building
#CLEAN=FORM::/[Bb]uilding/ 
# contains calculated,capability,capable,central,ceramic,certain,characteristic,character,choice,classical,clear,closed,cold,common,comparative,comparable
#CLEAN=FORM::/[Cc](ertain|hoice|lassical|omplicated|omputer|oncomitant|onsideration|onsiderable|ontext|onvenient|onventional|orresponding)/
# contains case
#CLEAN=FORM::/ [Cc]ases?/
# contains damage to
#CLEAN=FORM::/damage to/
# contains dangerous,decision, day,deep
#CLEAN=FORM::/[Dd](ecision|eeper|escription|etail|iffusible|igital|irection|iscussion)/
# contains early,easy,embodiment, empirical
CLEAN=FORM::/[Ee](mbodiment|xcellent)/
# contains etc.
CLEAN=FORM::/etc\./
# contains etc.
#CLEAN=FORM::/Inc\./
# starts with inch
CLEAN=FORM::/^inch/
# starts with other
CLEAN=FORM::/^other/
# starts with external
#CLEAN=FORM::/^external/
# contains other
CLEAN=FORM::/ other/
# contains facile
CLEAN=FORM::/[Ff](igure|ollowing|oregoing|urther)/
# starts with flow
#CLEAN=FORM::/^[Ff]low/
# starts with form
#CLEAN=FORM::/^[Ff]orms? /
# contains general
CLEAN=FORM::/[Gg](reater)/
# contains gram of
CLEAN=FORM::/^[Gg]rams? of/
# contains half of
#CLEAN=FORM::/[Hh]alf of/
# contains high
#CLEAN=FORM::/[Hh](olding|ollow)/
# contains ideal
CLEAN=FORM::/[Ii]llustration/
# starts with l
CLEAN=FORM::/^l /
# starts with II
CLEAN=FORM::/^(II|ii)/ 
# contains knowledge,
#CLEAN=FORM::/[Kk](nowledge)/
# contains lack of
#CLEAN=FORM::/[Ll]ack of/
# contains large
#CLEAN=FORM::/[Ll](oading)/
# starts with less
#CLEAN=FORM::/^[Ll]ess/
# starts with long
#CLEAN=FORM::/^[Ll]ong/
# starts with loss
#CLEAN=FORM::/^[Ll](oss|ow|ower) /
# contains magnitude
CLEAN=FORM::/[Mm](any|ode |oles? )/
# starts with main
#CLEAN=FORM::/^[Mm]ain /
# starts with min
#CLEAN=FORM::/^[Mm](in|olar) /
# starts with mean
#CLEAN=FORM::/^[Mm]eans? /
# starts with moiety to
CLEAN=FORM::/^[Mm]oiety to /
# contains member of
#CLEAN=FORM::/[Mm]ember of/
# contains method of
#CLEAN=FORM::/[Mm]ethods? of/
# contains middle of
#CLEAN=FORM::/[Mm]iddle of/
# starts with more
CLEAN=FORM::/^[Mm]ore /
# starts with several
#CLEAN=FORM::/^several/
# starts with none
#CLEAN=FORM::/^[Nn]one?/
# contains narrow
CLEAN=FORM::/[Nn](oble |onlimiting|otable)/
# contains need of
#CLEAN=FORM::/[Nn]eed of/
# contains number of
#CLEAN=FORM::/[Nn]umbers? of/
# contains objective
CLEAN=FORM::/[Oo](ngoing|nly|vernight|verview)/
# starts with object of
#CLEAN=FORM::/^[Oo](bjects?|bservations?) of/
# starts with other
CLEAN=FORM::/^[Oo]ther/
# contains page
CLEAN=FORM::/[Pp](ages?|atent|receding|remier|roceeding|urposes?)/
# starts with part of
#CLEAN=FORM::/^[Pp](arts?|assages?|ercentages?|ercents?|eriods?|ieces?|laces?|lurality|ortions?|roducts?) of/
# starts with product to
CLEAN=FORM::/^[Pp]roducts? to/
# contains quality
CLEAN=FORM::/[Qq](uasi)/
# contains random
#CLEAN=FORM::/[Rr](andom)/
# starts with rare, rate
CLEAN=FORM::/^[Rr](atios?|eference|efined|egard|eview)/
# contains respect to
CLEAN=FORM::/[Rr](espect) to/
# contains sale, same
CLEAN=FORM::/[Ss](ales?|ame|chematic|ee |hows|uch|uitable)/
# contains section of
#CLEAN=FORM::/[Ss](ections?) of/
# contains section xviii  
CLEAN=FORM::/[Ss]ection xviii/
# starts with sensation to
CLEAN=FORM::/^[Ss]ensations? to/
# starts with spatial, solid, special
#CLEAN=FORM::/^[Ss](tudy|upply) /
# starts with technical 
#CLEAN=FORM::/^[Tt](endency|erms?|he|heoretical|hick|housa) /
# contains undesirable  
CLEAN=FORM::/[Uu](pper) /
# starts with agt, atg ...  
CLEAN=FORM::/^(agt|atg|tgc|tcc|tgt|tta|gtt|ggt|gact|gtc|cca|cct|cgc|cga|ctg|CTG)/
# contains (1977))  
CLEAN=FORM::/^\([0-9][0-9]*\)\)/
# starts with (2S)  
CLEAN=FORM::/^\([0-9][0-9]*[A-Za-z]\)$/
# end by (s)  
CLEAN=FORM::/.*\([Ss]\)$/
# contains 5'  
CLEAN=FORM::/[0-9]\'/
# starts with  
CLEAN=FORM::/^Ph /
# methodsResultsDiscussionAcknowledgementsReferencesThe
CLEAN=FORM::/^[A-Za-z]{25,}/
#starts with against or about
CLEAN=FORM::/^(against|about)/
# contains only vitro, vivo
CLEAN=FORM::/^(vitro |vivo )/