bio 2.0.4 → 2.0.5

data/lib/bio/db/embl/uniprotkb.rb

@@ -530,22 +530,43 @@ class UniProtKB < EMBLDB
   # http://br.expasy.org/sprot/userman.html#OH_line
   def oh
     unless @data['OH']
-      @data['OH'] = fetch('OH').split("\. ").map {|x|
-        if x =~ /NCBI_TaxID=(\d+);/
-          taxid = $1
-        else
-          raise ArgumentError, ["Error: Invalid OH line format (#{self.entry_id}):",
-                                $!, "\n", get('OH'), "\n"].join
-          
-        end
-        if x =~ /NCBI_TaxID=\d+; (.+)/ 
-          host_name = $1
-          host_name.sub!(/\.$/, '')
-        else
-          host_name = nil
+      oh = []
+      a = fetch('OH').split(/(NCBI\_TaxID\=)(\d+)(\;)/)
+      t = catch :error do
+        taxid = nil
+        host_name = nil
+        while x = a.shift
+          x = x.to_s.strip
+          case x
+          when ''
+            next
+          when 'NCBI_TaxID='
+            if taxid then
+              oh.push({'NCBI_TaxID' => taxid, 'HostName' => host_name})
+              taxid = nil
+              host_name = nil
+            end
+            taxid = a.shift
+            throw :error, :missing_semicolon if a.shift != ';'
+          else
+            throw :error, :missing_taxid if host_name
+            host_name = x
+            host_name.sub!(/\.\z/, '')
+          end
+        end #while x...
+        if taxid then
+          oh.push({'NCBI_TaxID' => taxid, 'HostName' => host_name})
+        elsif host_name then
+          throw :error, :missing_taxid_last
         end
-        {'NCBI_TaxID' => taxid, 'HostName' => host_name}
-      }
+        nil
+      end #t = catch...
+      if t then
+        raise ArgumentError,
+              ["Error: Invalid OH line format (#{self.entry_id}):",
+               $!, "\n", get('OH'), "\n"].join
+      end
+      @data['OH'] = oh
     end
     @data['OH']
   end
@@ -922,6 +943,7 @@ class UniProtKB < EMBLDB
 
 
   def cc_alternative_products(data)
+    return nil unless data
     ap = data.join('')
     return ap unless ap
 
@@ -960,6 +982,7 @@ class UniProtKB < EMBLDB
 
 
   def cc_biophysiochemical_properties(data)
+    return nil unless data
     data = data[0]
 
     hash = {'Absorption' => {}, 
@@ -995,6 +1018,7 @@ class UniProtKB < EMBLDB
 
 
   def cc_caution(data)
+    return nil unless data
     data.join('')
   end
   private :cc_caution
@@ -1004,6 +1028,7 @@ class UniProtKB < EMBLDB
   #
   #   CC       P46527:CDKN1B; NbExp=1; IntAct=EBI-359815, EBI-519280;
   def cc_interaction(data)
+    return nil unless data
     str = data.join('')
     it = str.scan(/(.+?); NbExp=(.+?); IntAct=(.+?);/)
     it.map {|ent|
@@ -1059,6 +1084,7 @@ class UniProtKB < EMBLDB
 
 
   def cc_pathway(data)
+    return nil unless data
     data.map {|x| x.sub(/\.$/, '') }.map {|x|
       x.split(/; | and |: /)
     }[0]
@@ -1067,6 +1093,7 @@ class UniProtKB < EMBLDB
 
 
   def cc_rna_editing(data)
+    return nil unless data
     data = data.join('')
     entry = {'Modified_positions' => [], 'Note' => ""}
     if data =~ /Modified_positions=(.+?)(\.|;)/
@@ -1083,6 +1110,7 @@ class UniProtKB < EMBLDB
 
 
   def cc_subcellular_location(data)
+    return nil unless data
     data.map {|x| 
       x.split('. ').map {|y| 
         y.split('; ').map {|z| 
@@ -1101,6 +1129,7 @@ class UniProtKB < EMBLDB
   #++
 
   def cc_web_resource(data)
+    return nil unless data
     data.map {|x|
       entry = {'Name' => nil, 'Note' => nil, 'URL' => nil}
       x.split(';').each do |y|
@@ -1235,10 +1264,14 @@ class UniProtKB < EMBLDB
     begin
       ftlines.each do |line|
         if /^FT +([^\s]+) +(([^\s]+)\:)?([\<\?]?[0-9]+|\?)(?:\.\.([\>\?]?[0-9]+|\?))?\s*$/ =~ line
-          cur_ft = [$1.to_s,        # Feature Name
-                    "#{$2}#{$4}",   # From
-                    $5.to_s,        # To
-                    []              # Qualifiers
+          f_name = $1.to_s
+          f_from = "#{$2}#{$4}"
+          f_to = $5.to_s
+          f_to = f_from if f_to.empty?
+          cur_ft = [f_name,   # Feature Name
+                    f_from,   # From
+                    f_to,     # To
+                    []        # Qualifiers
                    ]
           table.push cur_ft
           cont = false

data/lib/bio/version.rb

@@ -10,7 +10,7 @@
 module Bio
 
   # BioRuby version (Array containing Integer)
-  BIORUBY_VERSION = [2, 0, 4].extend(Comparable).freeze
+  BIORUBY_VERSION = [2, 0, 5].extend(Comparable).freeze
 
   # Extra version specifier (String or nil).
   # Existance of the value indicates development version.

data/test/data/uniprot/P03589.uniprot

@@ -0,0 +1,127 @@
+ID   1A_AMVLE                Reviewed;        1126 AA.
+AC   P03589;
+DT   21-JUL-1986, integrated into UniProtKB/Swiss-Prot.
+DT   21-JUL-1986, sequence version 1.
+DT   22-FEB-2023, entry version 78.
+DE   RecName: Full=Replication protein 1a;
+DE   Includes:
+DE     RecName: Full=ATP-dependent helicase;
+DE              EC=3.6.4.-;
+DE   Includes:
+DE     RecName: Full=Methyltransferase;
+DE              EC=2.1.1.-;
+GN   ORFNames=ORF1a;
+OS   Alfalfa mosaic virus (strain 425 / isolate Leiden).
+OC   Viruses; Riboviria; Orthornavirae; Kitrinoviricota; Alsuviricetes;
+OC   Martellivirales; Bromoviridae; Alfamovirus.
+OX   NCBI_TaxID=12322;
+OH   NCBI_TaxID=4045; Apium graveolens (Celery).
+OH   NCBI_TaxID=83862; Astragalus glycyphyllos (Wild liquorice).
+OH   NCBI_TaxID=4072; Capsicum annuum (Capsicum pepper).
+OH   NCBI_TaxID=41386; Caryopteris incana.
+OH   NCBI_TaxID=3827; Cicer arietinum (Chickpea) (Garbanzo).
+OH   NCBI_TaxID=3847; Glycine max (Soybean) (Glycine hispida).
+OH   NCBI_TaxID=35936; Lablab purpureus (Hyacinth bean) (Dolichos lablab).
+OH   NCBI_TaxID=4236; Lactuca sativa (Garden lettuce).
+OH   NCBI_TaxID=3864; Lens culinaris (Lentil) (Cicer lens).
+OH   NCBI_TaxID=3869; Lupinus.
+OH   NCBI_TaxID=145753; Malva parviflora (Little mallow) (Cheeseweed mallow).
+OH   NCBI_TaxID=3879; Medicago sativa (Alfalfa).
+OH   NCBI_TaxID=4097; Nicotiana tabacum (Common tobacco).
+OH   NCBI_TaxID=3885; Phaseolus vulgaris (Kidney bean) (French bean).
+OH   NCBI_TaxID=23113; Philadelphus.
+OH   NCBI_TaxID=3888; Pisum sativum (Garden pea).
+OH   NCBI_TaxID=4081; Solanum lycopersicum (Tomato) (Lycopersicon esculentum).
+OH   NCBI_TaxID=4113; Solanum tuberosum (Potato).
+OH   NCBI_TaxID=157662; Teramnus repens.
+OH   NCBI_TaxID=60916; Trifolium incarnatum (Crimson clover).
+OH   NCBI_TaxID=85293; Viburnum opulus (High-bush cranberry).
+OH   NCBI_TaxID=3916; Vigna radiata var. radiata (Mung bean) (Phaseolus aureus).
+OH   NCBI_TaxID=3917; Vigna unguiculata (Cowpea).
+RN   [1]
+RP   NUCLEOTIDE SEQUENCE [GENOMIC RNA].
+RX   PubMed=6298738; DOI=10.1093/nar/11.5.1253;
+RA   Cornelissen B.J.C., Brederode F.T., Moormann R.J.M., Bol J.F.;
+RT   "Complete nucleotide sequence of alfalfa mosaic virus RNA 1.";
+RL   Nucleic Acids Res. 11:1253-1265(1983).
+CC   -!- FUNCTION: Involved in the virus replication. Contains a helicase domain
+CC       and a methyltransferase domain. The methyltransferase domain is
+CC       probably involved in viral RNA capping. Involved in the formation of ER
+CC       membrane spherular invaginations in which RNA replication complexes
+CC       form (By similarity). {ECO:0000250}.
+CC   -!- SUBUNIT: Interacts with RNA-directed RNA polymerase 2a. {ECO:0000250}.
+CC   -!- SUBCELLULAR LOCATION: Host endoplasmic reticulum membrane
+CC       {ECO:0000250}; Peripheral membrane protein {ECO:0000250}.
+CC   -!- SIMILARITY: Belongs to the bromoviridae replication protein 1a family.
+CC       {ECO:0000305}.
+CC   ---------------------------------------------------------------------------
+CC   Copyrighted by the UniProt Consortium, see https://www.uniprot.org/terms
+CC   Distributed under the Creative Commons Attribution (CC BY 4.0) License
+CC   ---------------------------------------------------------------------------
+DR   EMBL; L00163; AAA46289.1; -; Genomic_RNA.
+DR   PIR; A04197; WMFM12.
+DR   RefSeq; NP_041192.1; NC_001495.1.
+DR   SMR; P03589; -.
+DR   GeneID; 962667; -.
+DR   KEGG; vg:962667; -.
+DR   Proteomes; UP000000358; Genome.
+DR   GO; GO:0044167; C:host cell endoplasmic reticulum membrane; IEA:UniProtKB-SubCell.
+DR   GO; GO:0016020; C:membrane; IEA:UniProtKB-KW.
+DR   GO; GO:0005524; F:ATP binding; IEA:UniProtKB-KW.
+DR   GO; GO:0004386; F:helicase activity; IEA:UniProtKB-KW.
+DR   GO; GO:0016787; F:hydrolase activity; IEA:UniProtKB-KW.
+DR   GO; GO:0008174; F:mRNA methyltransferase activity; IEA:InterPro.
+DR   GO; GO:0003723; F:RNA binding; IEA:InterPro.
+DR   GO; GO:0006396; P:RNA processing; IEA:InterPro.
+DR   Gene3D; 3.40.50.300; P-loop containing nucleotide triphosphate hydrolases; 2.
+DR   InterPro; IPR027351; (+)RNA_virus_helicase_core_dom.
+DR   InterPro; IPR002588; Alphavirus-like_MT_dom.
+DR   InterPro; IPR027417; P-loop_NTPase.
+DR   Pfam; PF01443; Viral_helicase1; 1.
+DR   Pfam; PF01660; Vmethyltransf; 1.
+DR   SUPFAM; SSF52540; P-loop containing nucleoside triphosphate hydrolases; 1.
+DR   PROSITE; PS51743; ALPHAVIRUS_MT; 1.
+DR   PROSITE; PS51657; PSRV_HELICASE; 1.
+PE   3: Inferred from homology;
+KW   ATP-binding; Helicase; Host endoplasmic reticulum; Host membrane;
+KW   Hydrolase; Membrane; Methyltransferase; Nucleotide-binding;
+KW   Reference proteome; Transferase.
+FT   CHAIN           1..1126
+FT                   /note="Replication protein 1a"
+FT                   /id="PRO_0000083254"
+FT   DOMAIN          90..278
+FT                   /note="Alphavirus-like MT"
+FT                   /evidence="ECO:0000255|PROSITE-ProRule:PRU01079"
+FT   DOMAIN          806..963
+FT                   /note="(+)RNA virus helicase ATP-binding"
+FT   DOMAIN          964..1125
+FT                   /note="(+)RNA virus helicase C-terminal"
+FT   REGION          69..406
+FT                   /note="Methyltransferase"
+FT   REGION          834..1094
+FT                   /note="ATP-dependent helicase"
+FT   BINDING         838..845
+FT                   /ligand="ATP"
+FT                   /ligand_id="ChEBI:CHEBI:30616"
+FT                   /evidence="ECO:0000255"
+SQ   SEQUENCE   1126 AA;  125828 MW;  BF5A8019B47D4CBF CRC64;
+     MNADAQSTDA SLSMREPLSH ASIQEMLRRV VEKQAADDTT AIGKVFSEAG RAYAQDALPS
+     DKGEVLKISF SLDATQQNIL RANFPGRRTV FSNSSSSSHC FAAAHRLLET DFVYRCFGNT
+     VDSIIDLGGN FVSHMKVKRH NVHCCCPILD ARDGARLTER ILSLKSYVRK HPEIVGEADY
+     CMDTFQKCSR RADYAFAIHS TSDLDVGELA CSLDQKGVMK FICTMMVDAD MLIHNEGEIP
+     NFNVRWEIDR KKDLIHFDFI DEPNLGYSHR FSLLKHYLTY NAVDLGHAAY RIERKQDFGG
+     VMVIDLTYSL GFVPKMPHSN GRSCAWYNRV KGQMVVHTVN EGYYHHSYQT AVRRKVLVDK
+     KVLTRVTEVA FRQFRPNADA HSAIQSIATM LSSSTNHTII GGVTLISGKP LSPDDYIPVA
+     TTIYYRVKKL YNAIPEMLSL LDKGERLSTD AVLKGSEGPM WYSGPTFLSA LDKVNVPGDF
+     VAKALLSLPK RDLKSLFSRS ATSHSERTPV RDESPIRCTD GVFYPIRMLL KCLGSDKFES
+     VTITDPRSNT ETTVDLYQSF QKKIETVFSF ILGKIDGPSP LISDPVYFQS LEDVYYAEWH
+     QGNAIDASNY ARTLLDDIRK QKEESLKAKA KEVEDAQKLN RAILQVHAYL EAHPDGGKIE
+     GLGLSSQFIA KIPELAIPTP KPLPEFEKNA ETGEILRINP HSDAILEAID YLKSTSANSI
+     ITLNKLGDHC QWTTKGLDVV WAGDDKRRAF IPKKNTWVGP TARSYPLAKY ERAMSKDGYV
+     TLRWDGEVLD ANCVRSLSQY EIVFVDQSCV FASAEAIIPS LEKALGLEAH FSVTIVDGVA
+     GCGKTTNIKQ IARSSGRDVD LILTSNRSSA DELKETIDCS PLTKLHYIRT CDSYLMSASA
+     VKAQRLIFDE CFLQHAGLVY AAATLAGCSE VIGFGDTEQI PFVSRNPSFV FRHHKLTGKV
+     ERKLITWRSP ADATYCLEKY FYKNKKPVKT NSRVLRSIEV VPINSPVSVE RNTNALYLCH
+     TQAEKAVLKA QTHLKGCDNI FTTHEAQGKT FDNVYFCRLT RTSTSLATGR DPINGPCNGL
+     VALSRHKKTF KYFTIAHDSD DVIYNACRDA GNTDDSILAR SYNHNF
+//

data/test/data/uniprot/P49144.uniprot

@@ -0,0 +1,232 @@
+ID   5HT1B_RABIT             Reviewed;         390 AA.
+AC   P49144;
+DT   01-FEB-1996, integrated into UniProtKB/Swiss-Prot.
+DT   01-FEB-1996, sequence version 1.
+DT   22-FEB-2023, entry version 127.
+DE   RecName: Full=5-hydroxytryptamine receptor 1B;
+DE            Short=5-HT-1B;
+DE            Short=5-HT1B;
+DE   AltName: Full=Serotonin 1D beta receptor;
+DE            Short=5-HT-1D-beta;
+DE   AltName: Full=Serotonin receptor 1B;
+GN   Name=HTR1B;
+OS   Oryctolagus cuniculus (Rabbit).
+OC   Eukaryota; Metazoa; Chordata; Craniata; Vertebrata; Euteleostomi; Mammalia;
+OC   Eutheria; Euarchontoglires; Glires; Lagomorpha; Leporidae; Oryctolagus.
+OX   NCBI_TaxID=9986;
+RN   [1]
+RP   NUCLEOTIDE SEQUENCE [GENOMIC DNA].
+RC   TISSUE=Liver;
+RX   PubMed=8543023; DOI=10.1016/0014-5793(95)01308-3;
+RA   Harwood G.S., Lockyer M., Giles H., Fairweather N.;
+RT   "Cloning and characterisation of the rabbit 5-HT1D alpha and 5-HT1D beta
+RT   receptors.";
+RL   FEBS Lett. 377:73-76(1995).
+RN   [2]
+RP   NUCLEOTIDE SEQUENCE.
+RC   STRAIN=New Zealand white; TISSUE=Saphenous vein;
+RA   Wurch T., Cathala C., Palmer C., Valentin J.P., John G., Colpaert F.C.,
+RA   Pauwels P.J.;
+RT   "Molecular cloning and identification of a rabbit saphenous vein 5-HT 1DB
+RT   receptor gene.";
+RL   Neurosci. Res. Commun. 18:155-162(1996).
+RN   [3]
+RP   NUCLEOTIDE SEQUENCE.
+RC   STRAIN=New Zealand white;
+RX   PubMed=8878052; DOI=10.1007/bf00171053;
+RA   Bard J.A., Kucharewicz S.A., Zgombick J.M., Weinshank R.L., Branchek T.A.,
+RA   Cohen M.L.;
+RT   "Differences in ligand binding profiles between cloned rabbit and human 5-
+RT   HT1D alpha and 5-HT1D beta receptors: ketanserin and methiothepin
+RT   distinguish rabbit 5-HT1D receptor subtypes.";
+RL   Naunyn Schmiedebergs Arch. Pharmacol. 354:237-244(1996).
+CC   -!- FUNCTION: G-protein coupled receptor for 5-hydroxytryptamine
+CC       (serotonin). Also functions as a receptor for various alkaloids and
+CC       psychoactive substances. Ligand binding causes a conformation change
+CC       that triggers signaling via guanine nucleotide-binding proteins (G
+CC       proteins) and modulates the activity of down-stream effectors, such as
+CC       adenylate cyclase. Signaling inhibits adenylate cyclase activity.
+CC       Arrestin family members inhibit signaling via G proteins and mediate
+CC       activation of alternative signaling pathways. Regulates the release of
+CC       5-hydroxytryptamine, dopamine and acetylcholine in the brain, and
+CC       thereby affects neural activity, nociceptive processing, pain
+CC       perception, mood and behavior. Besides, plays a role in
+CC       vasoconstriction of cerebral arteries.
+CC   -!- SUBUNIT: Homodimer. Heterodimer with HTR1D (By similarity).
+CC       {ECO:0000250}.
+CC   -!- SUBCELLULAR LOCATION: Cell membrane; Multi-pass membrane protein.
+CC   -!- DOMAIN: Ligands are bound in a hydrophobic pocket formed by the
+CC       transmembrane helices. {ECO:0000250}.
+CC   -!- PTM: Phosphorylated. {ECO:0000250}.
+CC   -!- PTM: Palmitoylated. {ECO:0000250}.
+CC   -!- MISCELLANEOUS: A residue in the 7th transmembrane region ('Thr-355' in
+CC       human, 'Asn-351' in mouse and rat) is important for species-specific
+CC       sensitivity to various agonists. {ECO:0000250}.
+CC   -!- SIMILARITY: Belongs to the G-protein coupled receptor 1 family.
+CC       {ECO:0000255|PROSITE-ProRule:PRU00521}.
+CC   ---------------------------------------------------------------------------
+CC   Copyrighted by the UniProt Consortium, see https://www.uniprot.org/terms
+CC   Distributed under the Creative Commons Attribution (CC BY 4.0) License
+CC   ---------------------------------------------------------------------------
+DR   EMBL; Z50163; CAA90531.1; -; Genomic_DNA.
+DR   EMBL; X89731; CAA61883.1; -; mRNA.
+DR   EMBL; U60826; AAB58467.1; -; Genomic_DNA.
+DR   PIR; S58126; S58126.
+DR   PIR; S68422; S68422.
+DR   RefSeq; NP_001076259.1; NM_001082790.1.
+DR   AlphaFoldDB; P49144; -.
+DR   SMR; P49144; -.
+DR   STRING; 9986.ENSOCUP00000016295; -.
+DR   BindingDB; P49144; -.
+DR   ChEMBL; CHEMBL5717; -.
+DR   GlyCosmos; P49144; 2 sites, No reported glycans.
+DR   GeneID; 100009594; -.
+DR   KEGG; ocu:100009594; -.
+DR   CTD; 3351; -.
+DR   eggNOG; KOG3656; Eukaryota.
+DR   InParanoid; P49144; -.
+DR   OrthoDB; 2999405at2759; -.
+DR   TreeFam; TF316350; -.
+DR   PRO; PR:P49144; -.
+DR   Proteomes; UP000001811; Unplaced.
+DR   GO; GO:0005886; C:plasma membrane; ISS:UniProtKB.
+DR   GO; GO:0045202; C:synapse; IEA:GOC.
+DR   GO; GO:0004993; F:G protein-coupled serotonin receptor activity; ISS:UniProtKB.
+DR   GO; GO:0007198; P:adenylate cyclase-inhibiting serotonin receptor signaling pathway; ISS:UniProtKB.
+DR   GO; GO:0007610; P:behavior; IEA:UniProtKB-KW.
+DR   GO; GO:0046849; P:bone remodeling; IEA:InterPro.
+DR   GO; GO:0071312; P:cellular response to alkaloid; ISS:UniProtKB.
+DR   GO; GO:0071466; P:cellular response to xenobiotic stimulus; ISS:UniProtKB.
+DR   GO; GO:0007268; P:chemical synaptic transmission; IEA:InterPro.
+DR   GO; GO:0014063; P:negative regulation of serotonin secretion; ISS:UniProtKB.
+DR   GO; GO:0050795; P:regulation of behavior; IEA:InterPro.
+DR   GO; GO:0042310; P:vasoconstriction; IEA:InterPro.
+DR   CDD; cd15333; 7tmA_5-HT1B_1D; 1.
+DR   Gene3D; 1.20.1070.10; Rhodopsin 7-helix transmembrane proteins; 1.
+DR   InterPro; IPR002147; 5HT1B_rcpt.
+DR   InterPro; IPR002231; 5HT_rcpt.
+DR   InterPro; IPR000276; GPCR_Rhodpsn.
+DR   InterPro; IPR017452; GPCR_Rhodpsn_7TM.
+DR   PANTHER; PTHR24248; ADRENERGIC RECEPTOR-RELATED G-PROTEIN COUPLED RECEPTOR; 1.
+DR   PANTHER; PTHR24248:SF66; OCTOPAMINE RECEPTOR BETA-3R; 1.
+DR   Pfam; PF00001; 7tm_1; 1.
+DR   PRINTS; PR00513; 5HT1BRECEPTR.
+DR   PRINTS; PR01101; 5HTRECEPTOR.
+DR   PRINTS; PR00237; GPCRRHODOPSN.
+DR   SMART; SM01381; 7TM_GPCR_Srsx; 1.
+DR   SUPFAM; SSF81321; Family A G protein-coupled receptor-like; 1.
+DR   PROSITE; PS00237; G_PROTEIN_RECEP_F1_1; 1.
+DR   PROSITE; PS50262; G_PROTEIN_RECEP_F1_2; 1.
+PE   2: Evidence at transcript level;
+KW   Behavior; Cell membrane; Disulfide bond; G-protein coupled receptor;
+KW   Glycoprotein; Lipoprotein; Membrane; Palmitate; Phosphoprotein; Receptor;
+KW   Reference proteome; Transducer; Transmembrane; Transmembrane helix.
+FT   CHAIN           1..390
+FT                   /note="5-hydroxytryptamine receptor 1B"
+FT                   /id="PRO_0000068920"
+FT   TOPO_DOM        1..49
+FT                   /note="Extracellular"
+FT                   /evidence="ECO:0000250"
+FT   TRANSMEM        50..75
+FT                   /note="Helical; Name=1"
+FT                   /evidence="ECO:0000250"
+FT   TOPO_DOM        76..84
+FT                   /note="Cytoplasmic"
+FT                   /evidence="ECO:0000250"
+FT   TRANSMEM        85..110
+FT                   /note="Helical; Name=2"
+FT                   /evidence="ECO:0000250"
+FT   TOPO_DOM        111..123
+FT                   /note="Extracellular"
+FT                   /evidence="ECO:0000250"
+FT   TRANSMEM        124..145
+FT                   /note="Helical; Name=3"
+FT                   /evidence="ECO:0000250"
+FT   TOPO_DOM        146..165
+FT                   /note="Cytoplasmic"
+FT                   /evidence="ECO:0000250"
+FT   TRANSMEM        166..187
+FT                   /note="Helical; Name=4"
+FT                   /evidence="ECO:0000250"
+FT   TOPO_DOM        188..205
+FT                   /note="Extracellular"
+FT                   /evidence="ECO:0000250"
+FT   TRANSMEM        206..228
+FT                   /note="Helical; Name=5"
+FT                   /evidence="ECO:0000250"
+FT   TOPO_DOM        229..315
+FT                   /note="Cytoplasmic"
+FT                   /evidence="ECO:0000250"
+FT   TRANSMEM        316..336
+FT                   /note="Helical; Name=6"
+FT                   /evidence="ECO:0000250"
+FT   TOPO_DOM        337..349
+FT                   /note="Extracellular"
+FT                   /evidence="ECO:0000250"
+FT   TRANSMEM        350..371
+FT                   /note="Helical; Name=7"
+FT                   /evidence="ECO:0000250"
+FT   TOPO_DOM        372..390
+FT                   /note="Cytoplasmic"
+FT                   /evidence="ECO:0000250"
+FT   REGION          260..282
+FT                   /note="Disordered"
+FT                   /evidence="ECO:0000256|SAM:MobiDB-lite"
+FT   MOTIF           146..148
+FT                   /note="DRY motif; important for ligand-induced conformation
+FT                   changes and signaling"
+FT                   /evidence="ECO:0000250"
+FT   MOTIF           365..369
+FT                   /note="NPxxY motif; important for ligand-induced
+FT                   conformation changes and signaling"
+FT                   /evidence="ECO:0000250"
+FT   BINDING         129
+FT                   /ligand="ergotamine"
+FT                   /ligand_id="ChEBI:CHEBI:190463"
+FT                   /ligand_note="agonist"
+FT                   /evidence="ECO:0000250|UniProtKB:P28222"
+FT   BINDING         134
+FT                   /ligand="ergotamine"
+FT                   /ligand_id="ChEBI:CHEBI:190463"
+FT                   /ligand_note="agonist"
+FT                   /evidence="ECO:0000250|UniProtKB:P28222"
+FT   BINDING         201
+FT                   /ligand="ergotamine"
+FT                   /ligand_id="ChEBI:CHEBI:190463"
+FT                   /ligand_note="agonist"
+FT                   /evidence="ECO:0000250|UniProtKB:P28222"
+FT   SITE            355
+FT                   /note="Important for species-specific agonist sensitivity"
+FT                   /evidence="ECO:0000250"
+FT   LIPID           388
+FT                   /note="S-palmitoyl cysteine"
+FT                   /evidence="ECO:0000255"
+FT   CARBOHYD        24
+FT                   /note="N-linked (GlcNAc...) asparagine"
+FT                   /evidence="ECO:0000255"
+FT   CARBOHYD        32
+FT                   /note="N-linked (GlcNAc...) asparagine"
+FT                   /evidence="ECO:0000255"
+FT   DISULFID        122..199
+FT                   /evidence="ECO:0000255|PROSITE-ProRule:PRU00521"
+FT   CONFLICT        5
+FT                   /note="G -> S (in Ref. 1 and 3)"
+FT                   /evidence="ECO:0000305"
+FT   CONFLICT        7
+FT                   /note="Q -> R (in Ref. 1 and 3)"
+FT                   /evidence="ECO:0000305"
+FT   CONFLICT        14
+FT                   /note="Missing (in Ref. 1 and 3)"
+FT                   /evidence="ECO:0000305"
+FT   CONFLICT        171
+FT                   /note="R -> A (in Ref. 1 and 3)"
+FT                   /evidence="ECO:0000305"
+SQ   SEQUENCE   390 AA;  43496 MW;  C22EBC077C6C897D CRC64;
+     MEEPGAQCAP PLAAGSQIAV PQANLSAAHS HNCSAEGYIY QDSIALPWKV LLVLLLALFT
+     LATTLSNAFV VATVYRTRKL HTPANYLIAS LAVTDLLVSI LVMPISTMYT VTGRWTLGQV
+     VCDLWLSSDI TCCTASIMHL CVIALDRYWA ITDAVEYSAK RTPKRAAIMI RLVWVFSICI
+     SLPPFFWRQA KAEEEVSECL VNTDHVLYTV YSTVGAFYLP TLLLIALYGR IYVEARSRIL
+     KQTPNRTGKR LTRAQLITDS PGSTTSVTSI NSRAPDVPSE SGSPVYVNQV KVRVSDALLE
+     KKKLMAARER KATKTLGIIL GVFIVCWLPF FIISLVMPIC KDACWFHQAI FDFFTWLGYV
+     NSLINPIIYT MSNEDFKQAF HKLIRFKCTS
+//

data/test/functional/bio/test_command.rb

@@ -187,8 +187,10 @@ module Bio
         return
       end
       assert_equal('', str.to_s.strip)
+      assert_equal('', err.to_s.strip)
       str, err = Bio::Command.query_command_open3(ary, @data)
       assert_equal(@sorted, str.to_s.strip.split(/\s+/))
+      assert_equal('', err.to_s.strip)
     end
   end #class FuncTestCommandQuery
 

data/test/network/bio/db/kegg/test_genes_hsa7422.rb

@@ -38,7 +38,8 @@ module Bio
                   "H01457  Diabetic retinopathy",
                   "H01459  Diabetic neuropathy",
                   "H01529  Avascular necrosis of femoral head",
-                  "H01709  Glucocorticoid-induced osteonecrosis"]
+                  "H01709  Glucocorticoid-induced osteonecrosis",
+                  "H02559  Microvascular complications of diabetes"]
 
       assert_equal(expected, @obj.diseases_as_strings)
     end
@@ -48,41 +49,54 @@ module Bio
                   "H01457"=>"Diabetic retinopathy",
                   "H01459"=>"Diabetic neuropathy",
                   "H01529"=>"Avascular necrosis of femoral head",
-                  "H01709"=>"Glucocorticoid-induced osteonecrosis"}
+                  "H01709"=>"Glucocorticoid-induced osteonecrosis",
+                  "H02559"=>"Microvascular complications of diabetes"}
       assert_equal(expected, @obj.diseases_as_hash)
     end
 
     def test_drug_targets_as_strings
       expected = ["Abicipar pegol: D11517",
-                  "Aflibercept: D09574",
-                  "Aflibercept beta: D10819",
-                  "Bevacizumab: D06409",
+                  "Aflibercept: D09574<JP/US>",
+                  "Aflibercept beta: D10819<JP>",
+                  "Bevacizumab: D06409<JP/US>",
                   "Bevasiranib sodium: D08874",
-                  "Brolucizumab: D11083",
-                  "Faricimab: D11516",
+                  "Brolucizumab: D11083<JP/US>",
+                  "Dilpacimab: D11642",
+                  "Emvododstat: D11890",
+                  "Faricimab: D11516<JP/US>",
                   "Navicixizumab: D11126",
                   "Pegaptanib: D05386",
-                  "Ranibizumab: D05697",
+                  "Ranibizumab: D05697<JP/US>",
+                  "Tarcocimab: D12507",
+                  "Tarcocimab tedromer: D12508",
                   "Vanucizumab: D11244"]
       assert_equal(expected, @obj.drug_targets_as_strings)
     end
 
     def test_networks_as_strings
-      expected = ["nt06114  PI3K signaling (virus)",
-                  "nt06124  Chemokine signaling (virus)",
+      expected = ["nt06124  Chemokine signaling (viruses)",
                   "nt06164  Kaposi sarcoma-associated herpesvirus (KSHV)",
-                  "nt06214  PI3K signaling",
                   "nt06219  JAK-STAT signaling",
                   "nt06224  CXCR signaling",
                   "nt06225  HIF-1 signaling",
+                  "nt06227  Nuclear receptor signaling",
                   "nt06262  Pancreatic cancer",
                   "nt06264  Renal cell carcinoma",
+                  "nt06360  Cushing syndrome",
+                  "nt06526  MAPK signaling",
+                  "nt06528  Calcium signaling",
+                  "nt06530  PI3K signaling",
                   "N00079  HIF-1 signaling pathway",
                   "N00080  Loss of VHL to HIF-1 signaling pathway",
                   "N00081  Mutation-inactivated VHL to HIF-1 signaling pathway",
                   "N00095  ERBB2-overexpression to EGF-Jak-STAT signaling pathway",
                   "N00157  KSHV vGPCR to GNB/G-ERK signaling pathway",
-                  "N00179  KSHV K1 to PI3K-NFKB signaling pathway"]
+                  "N00317  AhR signaling pathway",
+                  "N01412  Metals to HTF-1 signaling pathway",
+                  "N01592  GF-RTK-RAS-ERK signaling pathway",
+                  "N01641  RTK-PLCG-ITPR signaling pathway",
+                  "N01656  GF-RTK-PI3K signaling pathway",
+                  "N01658  GF-RTK-RAS-PI3K signaling pathway"]
       assert_equal(expected, @obj.networks_as_strings)
     end