Package TEES :: Package Utils :: Module DefaultSettings
[hide private]

Source Code for Module TEES.Utils.DefaultSettings

  1  import os 
  2   
  3  # The default DATAPATH for installing data and tools 
  4  DATAPATH=os.path.expanduser("~/TEES_DATA") 
  5   
  6  # Default locations for evaluators and their devel set gold data (relative to DATAPATH) 
  7  EVALUATOR = {} 
  8  EVALUATOR["GE"] = "BioNLP-ST_2011_genia_tools_rev1" 
  9  EVALUATOR["EPI"] = "BioNLP-ST_2011_EPI-eval-tools" 
 10  EVALUATOR["ID"] = "BioNLP-ST_2011_ID-eval-tools" 
 11  EVALUATOR["BB"] = "BioNLP-ST_2011_Bacteria_Biotopes_evaluation_software" 
 12  EVALUATOR["BI"] = "BioNLP-ST_2011_bacteria_interactions_evaluation_software" 
 13  EVALUATOR["REN"] = "BioNLP-ST_2011_bacteria_rename_evaluation_sofware" 
 14  EVALUATOR["CO"] = "CREvalPackage1.4" 
 15  EVALUATOR["GE09"] = "bionlp09_shared_task_evaluation_tools_v1" 
 16  # Gold data for evaluators 
 17  EVALUATOR["GE-gold"] = "BioNLP-ST_2011_genia_devel_data_rev1.tar.gz" 
 18  EVALUATOR["EPI-gold"] = "BioNLP-ST_2011_Epi_and_PTM_development_data_rev1.tar.gz" 
 19  EVALUATOR["ID-gold"] = "BioNLP-ST_2011_Infectious_Diseases_development_data_rev1.tar.gz" 
 20  EVALUATOR["BB-gold"] = "BioNLP-ST_2011_Bacteria_Biotopes_dev_data_rev1.tar.gz" 
 21  EVALUATOR["BI-gold"] = "BioNLP-ST_2011_bacteria_interactions_dev_data_rev1-remixed.tar.gz" 
 22  EVALUATOR["REN-gold"] = "BioNLP-ST_2011_bacteria_rename_dev_data.tar.gz" 
 23  EVALUATOR["CO-gold"] = "BioNLP-ST_2011_coreference_development_data.tar.gz" 
 24  EVALUATOR["GE09-gold"] = "bionlp09_shared_task_development_data_rev1_for_evaluator.tar.gz" 
 25   
 26   
 27  # A dictionary for installation URLs. If there is a problem with a  
 28  # download location, it is possible to override a URL in the "TEES_SETTINGS"  
 29  # local settings file. 
 30  URL = {} 
 31   
 32  # Models 
 33   
 34  URL["MODELS"] = "https://github.com/downloads/jbjorne/TEES/TEES-models-120719.tar.bz2" 
 35   
 36  # External Tools ############################################################## 
 37   
 38  URL["SVM_MULTICLASS_SOURCE"] = "http://download.joachims.org/svm_multiclass/current/svm_multiclass.tar.gz" 
 39  URL["SVM_MULTICLASS_LINUX"] = "http://download.joachims.org/svm_multiclass/current/svm_multiclass_linux.tar.gz" 
 40  URL["GENIA_SENTENCE_SPLITTER"] = "http://www.nactem.ac.uk/y-matsu/geniass/geniass-1.00.tar.gz" 
 41  URL["BANNER_SOURCE"] = "http://banner.svn.sourceforge.net/viewvc/banner/trunk/?view=tar" 
 42  URL["BANNER_COMPILED"] = "https://github.com/downloads/jbjorne/TEES/BANNER-svn-snapshot-120630.tar.gz" 
 43  URL["BLLIP_SOURCE"] = "https://github.com/dmcc/bllip-parser/zipball/master" 
 44  URL["STANFORD_PARSER"] = "http://nlp.stanford.edu/software/stanford-parser-2012-03-09.tgz" 
 45  RUBY_PATH = "ruby" # for GENIA Sentence Splitter 
 46   
 47  # Corpora ##################################################################### 
 48   
 49  # Preconverted 
 50  URL["BIONLP_11_CORPORA"] =  "https://github.com/downloads/jbjorne/TEES/BioNLP11-corpora-XML-120715.zip" 
 51  URL["BIONLP_09_CORPUS"] =  "https://github.com/downloads/jbjorne/TEES/BioNLP09-corpus-XML-120715.zip" 
 52  URL["DDI_11_CORPUS"] =  "https://github.com/downloads/jbjorne/TEES/DDI11-corpus-XML-120715.zip" 
 53   
 54  # BioNLP'11 
 55  urlBase = "http://weaver.nlplab.org/~bionlp-st/BioNLP-ST/downloads/files/" 
 56  # Main tasks 
 57  URL["GE_DEVEL"] = urlBase + "BioNLP-ST_2011_genia_devel_data_rev1.tar.gz"  
 58  URL["GE_TRAIN"] = urlBase + "BioNLP-ST_2011_genia_train_data_rev1.tar.gz" 
 59  URL["GE_TEST"] = urlBase + "BioNLP-ST_2011_genia_test_data.tar.gz" 
 60  URL["EPI_DEVEL"] = urlBase + "BioNLP-ST_2011_Epi_and_PTM_development_data_rev1.tar.gz"  
 61  URL["EPI_TRAIN"] = urlBase + "BioNLP-ST_2011_Epi_and_PTM_training_data_rev1.tar.gz" 
 62  URL["EPI_TEST"] = urlBase + "BioNLP-ST_2011_Epi_and_PTM_test_data.tar.gz" 
 63  URL["ID_DEVEL"] = urlBase + "BioNLP-ST_2011_Infectious_Diseases_development_data_rev1.tar.gz"  
 64  URL["ID_TRAIN"] = urlBase + "BioNLP-ST_2011_Infectious_Diseases_training_data_rev1.tar.gz" 
 65  URL["ID_TEST"] = urlBase + "BioNLP-ST_2011_Infectious_Diseases_test_data.tar.gz" 
 66  URL["BB_DEVEL"] = urlBase + "BioNLP-ST_2011_Bacteria_Biotopes_dev_data_rev1.tar.gz"  
 67  URL["BB_TRAIN"] = urlBase + "BioNLP-ST_2011_Bacteria_Biotopes_train_data_rev1.tar.gz" 
 68  URL["BB_TEST"] = urlBase + "BioNLP-ST_2011_Bacteria_Biotopes_test_data.tar.gz" 
 69  URL["BI_DEVEL"] = urlBase + "BioNLP-ST_2011_bacteria_interactions_dev_data_rev1.tar.gz"  
 70  URL["BI_TRAIN"] = urlBase + "BioNLP-ST_2011_bacteria_interactions_train_data_rev1.tar.gz" 
 71  URL["BI_TEST"] = urlBase + "BioNLP-ST_2011_bacteria_interactions_test_data.tar.gz" 
 72  # Supporting tasks 
 73  URL["REL_DEVEL"] = urlBase + "BioNLP-ST_2011_Entity_Relations_development_data.tar.gz"  
 74  URL["REL_TRAIN"] = urlBase + "BioNLP-ST_2011_Entity_Relations_training_data.tar.gz" 
 75  URL["REL_TEST"] = urlBase + "BioNLP-ST_2011_Entity_Relations_test_data.tar.gz" 
 76  URL["REN_DEVEL"] = urlBase + "BioNLP-ST_2011_bacteria_rename_dev_data.tar.gz"  
 77  URL["REN_TRAIN"] = urlBase + "BioNLP-ST_2011_bacteria_rename_train_data.tar.gz" 
 78  URL["REN_TEST"] = urlBase + "BioNLP-ST_2011_bacteria_rename_test_data.tar.gz" 
 79  URL["CO_DEVEL"] = urlBase + "BioNLP-ST_2011_coreference_development_data.tar.gz" 
 80  URL["CO_TRAIN"] = urlBase + "BioNLP-ST_2011_coreference_training_data_rev1.tar.gz" 
 81  URL["CO_TEST"] = urlBase + "BioNLP-ST_2011_coreference_test_data.tar.gz" 
 82  # BioNLP'11 Evaluators 
 83  URL["BIONLP11_EVALUATORS"] = "https://github.com/downloads/jbjorne/TEES/BioNLP-evaluators-120715.tar.gz" 
 84  URL["GE_EVALUATOR"] = urlBase + "BioNLP-ST_2011_genia_tools_rev1.tar.gz" 
 85  #URL["EPI_EVALUATOR"] = urlBase +  
 86  #URL["ID_EVALUATOR"] = urlBase +  
 87  URL["BB_EVALUATOR"] = urlBase + "BioNLP-ST_2011_Bacteria_Biotopes_evaluation_software_rev2.tar.gz" 
 88  URL["BI_EVALUATOR"] = urlBase + "BioNLP-ST_2011_bacteria_interactions_evaluation_software.tar.gz" 
 89  #URL["REN_EVALUATOR"] = "http://sites.google.com/site/bionlpst/home/bacteria-gene-renaming-rename/BioNLP-ST_2011_bacteria_rename_evaluation_sofware.tar.gz" 
 90  URL["CO_EVALUATOR"] = urlBase + "CREvalPackage1.6.tar.gz" 
 91  # BioNLP'11 Supporting resources 
 92  urlBase = "http://weaver.nlplab.org/~bionlp-st/BioNLP-ST/downloads/support-files/" 
 93  #GE 
 94  URL["GE_TRAIN_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_genia_train_data_rev1.tar.gz" 
 95  URL["GE_DEVEL_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_genia_devel_data_rev1.tar.gz" 
 96  URL["GE_TEST_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_genia_test_data.tar.gz" 
 97  URL["GE_TRAIN_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_genia_train_data_rev1.tar.gz"  
 98  URL["GE_DEVEL_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_genia_devel_data_rev1.tar.gz" 
 99  URL["GE_TEST_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_genia_test_data.tar.gz" 
100  #EPI 
101  URL["EPI_DEVEL_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_Epi_and_PTM_development_data.tar.gz" 
102  URL["EPI_TRAIN_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_Epi_and_PTM_training_data.tar.gz" 
103  URL["EPI_TEST_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_Epi_and_PTM_test_data.tar.gz" 
104  URL["EPI_DEVEL_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_Epi_and_PTM_development_data.tar.gz"  
105  URL["EPI_TRAIN_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_Epi_and_PTM_training_data.tar.gz" 
106  URL["EPI_TEST_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_Epi_and_PTM_test_data.tar.gz" 
107  #ID 
108  URL["ID_DEVEL_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_Infectious_Diseases_development_data.tar.gz" 
109  URL["ID_TRAIN_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_Infectious_Diseases_training_data.tar.gz" 
110  URL["ID_TEST_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_Infectious_Diseases_test_data.tar.gz" 
111  URL["ID_DEVEL_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_Infectious_Diseases_development_data.tar.gz"  
112  URL["ID_TRAIN_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_Infectious_Diseases_training_data.tar.gz" 
113  URL["ID_TEST_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_Infectious_Diseases_test_data.tar.gz" 
114  #BB 
115  URL["BB_DEVEL_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_Bacteria_Biotopes_dev_data_rev1.tar.gz" 
116  URL["BB_TRAIN_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_Bacteria_Biotopes_train_data_rev1.tar.gz" 
117  URL["BB_TEST_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_Bacteria_Biotopes_test_data.tar.gz" 
118  URL["BB_DEVEL_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_Bacteria_Biotopes_dev_data_rev1.tar.gz"  
119  URL["BB_TRAIN_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_Bacteria_Biotopes_train_data_rev1.tar.gz" 
120  URL["BB_TEST_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_Bacteria_Biotopes_test_data.tar.gz" 
121  #BI 
122  URL["BI_DEVEL_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_bacteria_interactions_dev_data_rev1.tar.gz" 
123  URL["BI_TRAIN_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_bacteria_interactions_train_data_rev1.tar.gz" 
124  URL["BI_TEST_TOKENS"] = urlBase + "Tokenised-BioNLP-ST_2011_bacteria_interactions_test_data.tar.gz" 
125  URL["BI_DEVEL_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_bacteria_interactions_dev_data_rev1.tar.gz"  
126  URL["BI_TRAIN_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_bacteria_interactions_train_data_rev1.tar.gz" 
127  URL["BI_TEST_McCC"] = urlBase + "McCC-parses-BioNLP-ST_2011_bacteria_interactions_test_data.tar.gz" 
128  # Parses not provided in the shared tasks 
129  #URL["TEES_PARSES"] = "/home/jari/biotext/TEES-parses-2.0/TEES-parses-120706.tar.gz" 
130  URL["TEES_PARSES"] = "https://github.com/downloads/jbjorne/TEES/TEES-parses-120706.tar.gz" 
131   
132   
133  # BioNLP'09 Shared Task 
134  urlBase = "http://www.nactem.ac.uk/tsujii/GENIA/SharedTask/dat/" 
135  URL["GE09_DEVEL"] = urlBase + "bionlp09_shared_task_development_data_rev1.tar.gz"  
136  URL["GE09_TRAIN"] = urlBase + "bionlp09_shared_task_training_data_rev2.tar.gz" 
137  URL["GE09_TEST"] = urlBase + "bionlp09_shared_task_test_data_without_gold_annotation.tar.gz" 
138  # BioNLP'09 Evaluator 
139  URL["GE09_EVALUATOR"] = urlBase + "bionlp09_shared_task_evaluation_tools_v1.tar.gz" 
140  # BioNLP'09 Shared Task parses 
141  urlBase = "http://www-tsujii.is.s.u-tokyo.ac.jp/GENIA/SharedTask/dat/" 
142  URL["GE09_DEVEL_ANALYSES"] = urlBase + "bionlp09_shared_task_development_analyses_rev2.tar.gz"  
143  URL["GE09_TRAIN_ANALYSES"] = urlBase + "bionlp09_shared_task_training_analyses_rev2.tar.gz" 
144  URL["GE09_TEST_ANALYSES"] = urlBase + "bionlp09_shared_task_test_analyses_rev2.tar.gz" 
145   
146   
147  # DDI 
148  urlBase = "http://labda.inf.uc3m.es/DDIExtraction2011/" 
149  URL["DDI_TRAIN_UNIFIED"] = urlBase + "DrugDDI_Unified.zip" 
150  URL["DDI_TRAIN_MTMX"] = urlBase + "DrugDDI_MTMX.zip" 
151  # If you have registered for the DDI Shared Task, insert the paths of your downloaded 
152  # test files in the following variables (in your local settings file) to have them  
153  # converted for use with TEES 
154  URL["DDI_TEST_UNIFIED"] = None 
155  URL["DDI_TEST_MTMX"] = None 
156   
157   
158  # Miscellaneous files 
159  URL["TEES_RESOURCES"] = "https://github.com/downloads/jbjorne/TEES/TEES-resources-120705.tar.gz" 
160  URL["DRUG_BANK_XML"] = "http://www.drugbank.ca/system/downloads/current/drugbank.xml.zip" 
161