rbbt 1.1.7 → 2.0.0

data/install_scripts/organisms/rgd.Rakefile

@@ -1,88 +0,0 @@
-require __FILE__.sub(/[^\/]*$/,'') + '../rake-include'
-
-$name = "Rattus norvegicus"
-
-
-$native_id = "RGD DB ID"
-
-$entrez2native = {
-  :tax => 10116,
-  :check => proc{|code| code.match(/^RGD/)},
-}
-
-$lexicon = {
-  :file => {
-    :url => "ftp://rgd.mcw.edu/pub/data_release/gene_association.rgd.gz",
-    :native => 1,
-    :extra => [2,9],
-    :exclude => proc{|l| !l.match(/^RGD/)}
-  },
-}
-
-$identifiers = {
-  :file => {
-    :url => "ftp://rgd.mcw.edu/pub/data_release/gene_association.rgd.gz",
-    :native => 1,
-    :extra => [],
-    :exclude => proc{|l| !l.match(/^RGD/)}
-  },
-  :biomart => {
-    :database => 'rnorvegicus_gene_ensembl',
-    :main =>  ['Entrez Gene ID' , "entrezgene"], 
-    :extra => [ 
-      ['Associated Gene Name' , "external_gene_id"], 
-      ['Protein ID' , "protein_id"] , 
-      ['UniProt/SwissProt ID' , "uniprot_swissprot"] , 
-      ['UniProt/SwissProt Accession' , "uniprot_swissprot_accession"] , 
-      ['RefSeq Protein ID' , "refseq_peptide"] , 
-      ['EMBL (Genbank) ID' , "embl"] , 
-
-      ['Affy rae230a', "affy_rae230a"],
-      ['Affy rae230b', "affy_rae230b"],
-      ['Affy RaGene', "affy_ragene_1_0_st_v1"],
-      ['Affy rat230 2', "affy_rat230_2"],
-      ['Affy RaEx', "affy_raex_1_0_st_v1"],
-      ['Affy rg u34a', "affy_rg_u34a"],
-      ['Affy rg u34b', "affy_rg_u34b"],
-      ['Affy rg u34c', "affy_rg_u34c"],
-      ['Affy rn u34', "affy_rn_u34"],
-      ['Affy rt u34', "affy_rt_u34"],
-      ['Agilent WholeGenome',"agilent_wholegenome" ],
-      ['Codelink ID ', "codelink"],
-
-
-    ],
-    :filter => [],
-  }
-}
-
-$go = {
-  :url => "ftp://rgd.mcw.edu/pub/data_release/gene_association.rgd.gz",
-  :exclude => proc{|l| !l.match(/^RGD/)},
-  :code => 1,
-  :go   => 4,
-  :pmid => 5,
-}
-
-$query = '(("mice"[TIAB] NOT Medline[SB]) OR "mice"[MeSH Terms] OR mouse[Text Word]) AND ((("proteins"[TIAB] NOT Medline[SB]) OR "proteins"[MeSH Terms] OR protein[Text Word]) OR (("genes"[TIAB] NOT Medline[SB]) OR "genes"[MeSH Terms] OR gene[Text Word]))  AND hasabstract[text] AND English[lang]'
-
-#{{{ Redefines
-
-module Open
-
-  class << self
-    alias_method :old_read, :read
-
-    def read(url, options = {})
-      data = old_read(url, options)
-
-      if url =~ /gene_association.rgd.gz/
-        return data.collect{|l| l.gsub(/^RGD\t/,"RGD\tRGD:")}.join("\n")
-      else
-        return data
-      end
-
-    end
-  end
-end
-

data/install_scripts/organisms/sgd.Rakefile

@@ -1,66 +0,0 @@
-require __FILE__.sub(/[^\/]*$/,'') + '../rake-include'
-
-$name = "Saccharomyces cerevisiae"
-
-
-$native_id = "SGD DB Id"
-
-$entrez2native = {
-  :tax =>  4932,
-  :fix => proc{|code| code.sub(/SGD:S0/,'S0') },
-  :check => proc{|code| code.match(/^S0/)},
-}
-
-$lexicon = {
-  :file => {
-    :url => "ftp://genome-ftp.stanford.edu/pub/yeast/data_download/chromosomal_feature/SGD_features.tab",
-    :native => 0,
-    :extra => [4,3,5]
-  },
-  :biomart => {
-    :database => 'scerevisiae_gene_ensembl',
-    :main => ['Entrez Gene ID', 'entrezgene'],
-    :extra => [ 
-      ['Interpro Description' , "interpro_description"], 
-    ],
-    :filter => [],
-  }
-
-}
-
-$identifiers = {
-  :file => {
-    :url => "ftp://genome-ftp.stanford.edu/pub/yeast/data_download/chromosomal_feature/SGD_features.tab",
-    :native => 0,
-    :extra => [],
-  },
-  :biomart => {
-    :database => 'scerevisiae_gene_ensembl',
-    :main => ['Entrez Gene ID', 'entrezgene'],
-    :extra => [ 
-      ['Associated Gene Name' , "external_gene_id"], 
-      ['Ensembl Gene ID', "ensembl_gene_id"  ],
-      ['Ensembl Protein ID', "ensembl_peptide_id"  ],
-      ['RefSeq Protein ID' , "refseq_peptide"] , 
-      ['UniProt/SwissProt ID' , "uniprot_swissprot"] , 
-      ['UniProt/SwissProt Accession' , "uniprot_swissprot_accession"] , 
-      ['Protein ID' , "protein_id"] , 
-      ['EMBL (Genbank) ID' , "embl"] , 
-      # Affymetrix
-      ['Affy yeast 2',"affy_yeast_2"],
-      ['Affy yg s98', "affy_yg_s98"],
-    ],
-    :filter => [],
-  }
-}
-
-$go = {
-  :url => "ftp://genome-ftp.stanford.edu/pub/yeast/data_download/literature_curation/gene_association.sgd.gz",
-  :code => 1,
-  :go   => 4,
-  :pmid => 5,
-}
-
-$query = '"saccharomyces cerevisiae"[All Fields] AND ((("proteins"[TIAB] NOT Medline[SB]) OR "proteins"[MeSH Terms] OR protein[Text Word]) OR (("genes"[TIAB] NOT Medline[SB]) OR "genes"[MeSH Terms] OR gene[Text Word]))  AND hasabstract[text] AND English[lang]'
-
-

data/install_scripts/organisms/tair.Rakefile

@@ -1,54 +0,0 @@
-require __FILE__.sub(/[^\/]*$/,'') + '../rake-include'
-
-$name = "Arabidopsis thaliana"
-
-
-$native_id = "TAIR Locus"
-
-$entrez2native = {
-  :tax =>3702,
-  :fix => proc{|code| code.sub(/^TAIR:/,'')},
-  :check => proc{|code| true },
-}
-
-$lexicon = {
-  :file => {
-    :url => "ftp://ftp.arabidopsis.org/home/tair/Genes/gene_aliases.20090313",
-    :native => 0,
-    :extra => [1,2],
-  },
-}
-
-$identifiers = {
-  :file => {
-    :url => "ftp://ftp.arabidopsis.org/home/tair/Genes/gene_aliases.20090313",
-    :native => 0,
-    :extra => [],
-  },
-  :biomart => {
-    :database => 'athaliana_eg_gene',
-    :main => ['TAIR Locus', 'tair_locus'],
-    :extra => [
-      ['Associated Gene Name' , "external_gene_id"] ,
-      ['Gramene Gene ID' , "ensembl_gene_id"] ,
-      ['RefSeq peptide' , "refseq_peptide"] ,
-      ['Unigene' , "unigene"] ,
-      ['Interpro ID' , "interpro"] ,
-  
-  
-    ],
-    :filter => ['with_tair_locus'], # This is needed as the filter is not with_mgi_id as was expected
-  }
-
-}
-
-$go = {
-  :url =>  "ftp://ftp.arabidopsis.org/home/tair/Ontologies/Gene_Ontology/ATH_GO_GOSLIM.txt",
-  :code => 0,
-  :go   => 5,
-  :pmid => 12,
-}
-
-$query = '("arabidopsis"[MeSH Terms] OR Arabidopsis[Text Word]) AND ((("proteins"[TIAB] NOT Medline[SB]) OR "proteins"[MeSH Terms] OR protein[Text Word]) OR (("genes"[TIAB] NOT Medline[SB]) OR "genes"[MeSH Terms] OR gene[Text Word]))'
-
-

data/install_scripts/organisms/worm.Rakefile

@@ -1,109 +0,0 @@
-require __FILE__.sub(/[^\/]*$/,'') + '../rake-include'
-
-$name = "Caenorhabditis elegans"
-
-
-$native_id = "WormBase ID"
-
-$entrez2native = {
-  :tax => 6239,
-  :fix => proc{|code| code.sub(/^WormBase:/,'')},
-  :check => proc{|code| code.match(/^WBGene/)},
-}
-
-$lexicon = {
-
-  :file =>{
-
-    :url => "ftp://ftp.wormbase.org/pub/wormbase/genomes/elegans/annotations/GO/current.txt.gz",
-    :native => 0,
-    :extra   => [1,2], 
-
-#    :url => "ftp://ftp.wormbase.org/pub/wormbase/genomes/elegans/annotations/gene_ids/current.gz",
-#    :native => 0,
-#    :extra => [2,3,4,5],
-  
-  },
-}
-
-
-$identifiers = {
-
-  :file =>{
-
-    :url => "ftp://ftp.wormbase.org/pub/wormbase/genomes/elegans/annotations/GO/current.txt.gz",
-    :native => 0,
-    :extra   => [1,2], 
-
-#    :url => "ftp://ftp.wormbase.org/pub/wormbase/genomes/elegans/annotations/gene_ids/current.gz",
-#    :native => 0,
-#    :extra => [2,3,4,5],
-  
-  },
-
-  :biomart => {
-    :database => 'celegans_gene_ensembl',
-    :main => ['Entrez Gene ID' , "entrezgene"], 
-    :extra => [ 
-      ['WormBase gene', "wormbase_gene"  ],
-      ['Associated Gene Name ', "external_gene_id"  ],
-      ['WormPep id', "wormpep_id"  ],
-      [ 'Ensembl Gene ID', "ensembl_gene_id"  ],
-      [ 'Ensembl Protein ID', "ensembl_peptide_id"  ],
-      [ 'Protein ID ', "protein_id"  ],
-      [ 'RefSeq Protein ID ', "refseq_peptide"  ],
-      [ 'Unigene ID ', "unigene"  ],
-      [ 'UniProt/SwissProt ID', "uniprot_swissprot"  ],
-      [ 'UniProt/SwissProt Accession', "uniprot_swissprot_accession"  ],
-      ['EMBL (Genbank) ID' , "embl"] , 
-    ],
-    :filter => [],
-  }
-}
-
-$go = {
-  :url => "ftp://ftp.wormbase.org/pub/wormbase/genomes/elegans/annotations/GO/current.txt.gz",
-  :code => 0,
-  :go   => 3,
-  :pmid => 3,
-}
-
-$query = '"caenorhabditis elegans"[MeSH Terms] OR Caenorhabditis elegans[Text Word]'
-##########################
-
-
-module Open
-
-  class << self
-    alias_method :old_read, :read
-
-    def read(url, options = {})
-      content = old_read(url, options)
-
-      if url =~ /GO/
-        return content.gsub(/.*:.*\((GO:\d+)\)/,'\1').gsub(/\nGO/,"|GO").
-                collect{|l|
-                  l = l.sub(/\|/,"\t")
-                  names, gos = l.chomp.split(/\t/)
-
-                  id, name, extra = names.split(/ /)
-                  extra = extra.gsub(/[()]/,'') if extra
-
-                  if gos
-                    gos.split(/\|/).collect{|go|
-                      [id, name, extra, go].join("\t")
-                    }.join("\n")
-                  else
-                    [id, name, extra].join("\t") + "\n"
-                  end
-                }.join("\n")
-      elsif url =~ /gene_ids/
-        return content.gsub(/,/,"\t")
-      else
-        return content
-      end
-
-    end
-  end
-end
-

data/install_scripts/wordlists/consonants

@@ -1,897 +0,0 @@
-bbr
-sts
-nds
-dgm
-ptl
-ptn
-bsc
-ntl
-nth
-ntm
-rbs
-rpt
-bst
-rdl
-nts
-nsh
-pts
-ccl
-rns
-mpl
-ngl
-rds
-ccr
-rst
-ckn
-ctr
-cts
-nct
-cch
-ndr
-dbl
-dcr
-dgr
-dsh
-dsm
-rth
-rdp
-rkb
-rkd
-rkh
-ngm
-ngs
-rkm
-rkp
-rkr
-rks
-rkt
-rld
-rmh
-rms
-rmw
-rsh
-ldb
-ldn
-thf
-ths
-nch
-stl
-tch
-ggl
-ght
-nkl
-stb
-stw
-ngd
-ngf
-ndh
-cht
-nks
-lps
-lds
-tst
-chs
-lst
-rsn
-rlp
-rlw
-ttl
-rls
-spr
-dth
-ldc
-ldf
-ldl
-lfr
-llf
-msb
-ngn
-llp
-lls
-ndb
-ndf
-ndj
-ndl
-ndm
-ndw
-ngb
-nst
-rgr
-ntr
-shb
-shf
-sps
-stf
-thd
-thh
-rdr
-bbl
-lfg
-lfh
-rbl
-rml
-rmt
-rts
-shc
-shd
-ngt
-shr
-shs
-rcr
-rfr
-rpr
-rsk
-rsp
-rwh
-lth
-ghs
-llb
-llk
-lsh
-lts
-stc
-ngh
-stm
-lch
-lwr
-ntn
-lks
-ltm
-ddl
-ffl
-nsc
-lkw
-ltz
-rch
-thr
-ssl
-ssn
-cks
-btr
-npl
-npr
-str
-shl
-nsk
-nsl
-nsm
-nsp
-ssf
-nwh
-pgr
-nwr
-pbr
-rcl
-psh
-pst
-psw
-rpl
-rsc
-rwr
-scl
-ntf
-nfl
-xpl
-ldm
-nfr
-ngr
-mpr
-rnk
-mpt
-nkn
-rnc
-rnp
-rnt
-rtl
-sks
-lft
-scr
-ltr
-mbr
-ggr
-nsw
-ppr
-ttr
-wns
-nbl
-nbr
-ckl
-ncl
-ncr
-rbr
-rfl
-nsv
-nss
-mbl
-chr
-mph
-mps
-fts
-thb
-thc
-thl
-thp
-pkn
-ppl
-chd
-ghn
-nsd
-nsf
-nsg
-ghb
-ghf
-ghg
-ghl
-ndt
-mbn
-mbs
-mbt
-rsd
-thw
-dst
-llg
-llh
-llt
-xtb
-xts
-nkf
-ckh
-ckw
-ctf
-ctl
-lsp
-ffs
-skm
-Tch
-mst
-chn
-rfb
-tsh
-tbr
-ftl
-ftn
-zzl
-rdf
-bgr
-bpl
-bsp
-btl
-ffr
-lph
-ghw
-ckb
-ckc
-ckm
-ckp
-ckr
-rmb
-rnl
-rnn
-ndp
-rkl
-msh
-rsm
-pch
-ffn
-lln
-chl
-sph
-llw
-spl
-Spr
-shn
-wsh
-wst
-ffb
-ftb
-fth
-ftw
-mnl
-rgh
-ndn
-sms
-gns
-gnt
-ntw
-thn
-pdr
-wbl
-wdr
-wfl
-wpl
-gfr
-gnp
-nkh
-xth
-llc
-rpn
-rps
-wls
-psk
-pwr
-rtf
-rtt
-rtw
-rtb
-rtc
-rtn
-phs
-lfl
-rfd
-rfs
-sch
-ptr
-Sch
-rnf
-tsm
-tsw
-ffm
-lpt
-lsk
-lcl
-lpl
-ltb
-ltc
-ltl
-ltp
-ltw
-ndc
-pph
-stn
-ftr
-ghc
-ghh
-ghr
-shm
-stp
-thm
-chm
-tts
-ngw
-phr
-chf
-dbr
-dsk
-psc
-spb
-ddr
-gnm
-lms
-llm
-mpk
-rrh
-nsb
-rtz
-sth
-ssm
-rtr
-std
-stg
-mpn
-tkn
-rph
-phl
-gsk
-gst
-lgr
-ngp
-psq
-nkt
-nns
-rkw
-rsl
-ssb
-ssk
-ssp
-ssw
-wnb
-ntb
-gms
-tcl
-tcr
-tdr
-tfl
-tgr
-tpl
-tsc
-tsk
-tsp
-ddb
-dds
-ffh
-pht
-pth
-rdn
-msk
-ckt
-wsb
-wsc
-wsf
-wsl
-wsm
-wsp
-wsr
-wss
-wsv
-wsw
-wts
-tzs
-dfl
-dsl
-skr
-shv
-ntg
-ntp
-rtg
-lkm
-sbr
-ssh
-sst
-swr
-lsm
-rlb
-tth
-McC
-lpr
-ckj
-dsp
-wbr
-wdl
-mpp
-pfr
-ftf
-Khm
-llj
-ckd
-lbr
-hnn
-hns
-rgl
-rnm
-shw
-nkb
-nkp
-nkw
-nnk
-rtm
-ctm
-ctn
-rct
-xpr
-xtr
-mpb
-mns
-rnb
-mdr
-lpf
-lpm
-ldr
-nck
-ndq
-gsh
-ndg
-ndk
-rdb
-rdc
-rdh
-rdt
-rdw
-rmf
-wks
-dch
-dph
-msm
-wth
-ldh
-lfb
-lfp
-lfw
-ndd
-ssr
-ngk
-nkg
-rlf
-rlh
-ssc
-lcr
-gwr
-tsb
-rsw
-rsq
-xgl
-ncs
-wch
-lkl
-lkt
-tpr
-ldw
-lmd
-rmn
-shp
-dgl
-lkn
-rrs
-xch
-xcl
-xcr
-lfs
-nsn
-lmm
-nkm
-mpm
-mbb
-mbw
-sdr
-bbs
-cst
-ggb
-ggc
-ggh
-ggn
-ggp
-ggs
-msd
-rkn
-vsk
-btf
-bts
-wnc
-wnf
-wng
-wnh
-wnp
-wnr
-wnt
-wnw
-ctw
-gtr
-scs
-sgr
-skw
-phn
-wsk
-xsw
-spn
-wds
-rnw
-ntd
-chw
-ckf
-gwh
-Chr
-mfl
-msc
-msp
-mpd
-bdr
-lml
-lmn
-mpf
-mpg
-skl
-skn
-dcl
-dsw
-ddh
-lbs
-lkh
-lld
-llr
-kcl
-ksh
-kst
-hms
-dfr
-kth
-tht
-bsl
-rdm
-mbp
-wgr
-rck
-nkr
-chc
-chh
-fst
-wkw
-ckg
-sht
-rcs
-rmc
-rmp
-fghbstr
-rdpl
-rdsw
-rksh
-rldl
-rlds
-rldw
-rthl
-rthw
-chts
-rdst
-ngst
-dths
-ndbr
-ndsc
-ndsh
-ndsl
-ndst
-ndsw
-ngsp
-tchc
-tchd
-thdr
-thst
-rmth
-shcl
-shst
-tchb
-tchf
-tchm
-tchs
-tcht
-ngth
-rstr
-ghtl
-ghts
-llkn
-llsp
-lshm
-stch
-lksw
-llfl
-nstr
-nsch
-nscr
-pstr
-rnst
-lfth
-wnst
-nchr
-nspl
-nssh
-ckst
-mphs
-thbr
-rchl
-chst
-wnsc
-wnsf
-wnsh
-wnsm
-wnsp
-nsgr
-ghbr
-ghtf
-ndth
-mbsc
-ghtr
-ghtw
-mbst
-mptr
-mpts
-nths
-nksg
-tchg
-rdsm
-bscr
-rscr
-ckbr
-rdsh
-rths
-rtsm
-rtsw
-ndpr
-ndtr
-ngsh
-tchp
-rksk
-rpsh
-llsh
-rtfr
-rtsl
-ckpr
-rtbr
-rtst
-lchr
-schn
-Schw
-ptwr
-mstr
-ltsh
-ndbl
-nskr
-rsch
-ghsh
-ndsm
-stpr
-ckcl
-thms
-cksh
-ghth
-ghtn
-mptl
-stgr
-stsc
-nkst
-ncts
-llst
-ttsb
-nthr
-tchw
-wnbr
-rthr
-ntbr
-mphl
-rchm
-rsts
-tspr
-tstr
-ffsh
-ffsp
-ffst
-ldsm
-phth
-wsfl
-wspr
-wssh
-wsst
-tzsc
-ghtc
-ghtd
-ghtg
-ghtm
-ghtt
-rchg
-nthl
-thpr
-lksh
-llwh
-dstr
-dsts
-rksm
-rshm
-tchl
-ncht
-lstr
-nspr
-cksm
-ghtb
-mbsk
-mpsh
-rksp
-tchk
-sthm
-mphr
-nchb
-ntsm
-lmsm
-nchm
-ghsc
-nckl
-tchh
-ndcl
-ndwr
-rdsc
-rshl
-rshn
-lthf
-rtth
-wths
-rspr
-ndch
-rshw
-rlfr
-ssbl
-llbl
-ngpl
-rthc
-cksb
-nchw
-ntst
-lmst
-rstb
-rsth
-rmst
-ldsp
-rpts
-nshr
-rthb
-rspl
-rthm
-rthq
-ggpl
-ggsh
-ghtp
-wngr
-wntr
-wstr
-nctl
-rtsp
-tsch
-pths
-thbl
-tthr
-chsh
-rtsh
-ftsm
-nksh
-wlsp
-sscr
-ssbr
-ssch
-rksc
-rnbr
-rnfl
-lths
-cksc
-rchw
-mscr
-rtwh
-shwh
-lfsk
-mpgr
-nksm
-cksk
-ckwh
-llfr
-kthr
-stpl
-stst
-mbpr
-mbsh
-rthp
-rthd
-dspr
-schw
-ckdr
-ckgr
-cksl
-cksp
-cktr
-llpl
-shtr
-rchb
-rchd
-rcht
-rmchrldsh
-rthwh
-chtsm
-ndscr
-ngspr
-tchcr
-tchst
-ngths
-llspr
-ndths
-mbscr
-tchbl
-ndthr
-rtswr
-stscr
-ffspr
-rdstr
-tzsch
-ghtcl
-ghtdr
-ghtsh
-ghtst
-ngthw
-ckstr
-ghsch
-rdscr
-rtstr
-rtthr
-lmstr
-rthsh
-wnstr
-tschm
-rkscr
-tchph
-rscht
-ststr
-rthpl
-ckscr
-rmstrtchstr
-tchphr