semtools 0.1.8 → 0.1.91

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a3f63cc6548a9938e31121d2018d1c1c477987007c5d253b5fa814a285bdb576
-  data.tar.gz: e1911d3157c3046590ca13bc86215d2260b4a8b2b1b25affa5c2673881036795
+  metadata.gz: c1748a4acbeed0c35e072aad9433498fe72cc70af5b2e42b27b246c306cc822b
+  data.tar.gz: 8c03e4a77c4daeea26691210f84d7c5924ee71e1199a5fe6fe21c67e9d297acb
 SHA512:
-  metadata.gz: 30c95df80957a4a35b6fea05b9552352f529d8e45c10f6b128924a3ce2ee5d90e92a1e9d5fe0016d25538147e12d3a9199c81222642c94cdd0eb3c89eea168ef
-  data.tar.gz: ddc9e600fd984e68d060b7be05adf27b3f20bb67e638d42acc4b9b156eedabfce20d6f588a03d1fbc2948fedbd80d498f1767c0e3f8ea03720fa0ca327b95f3c
+  metadata.gz: 86eb4530e233bb8ef0e863e7486f42550d37e927988a6c7a8815e487e3c11c25b8fea199a5362a7be109ee6cafd6f9323f9906185504f0c47746d0768584adbd
+  data.tar.gz: 0a11d9a0194325bd08f53342f250e798a3cce9eec1377148128a3b8df9a48efb88be625c08b4f36944ad729239537f4f343ad6ccd00bca8ba73b1fc10e828e77

data/README.md

@@ -1,5 +1,7 @@
 # Semtools
 
+DEPRECATED PROJECT. MIGRATED TO [python semtools](https://github.com/seoanezonjic/py_semtools)
+
 Welcome to your new gem! In this directory, you'll find the files you need to be able to package up your Ruby library into a gem. Put your Ruby code in the file `lib/semtools`. To experiment with that code, run `bin/console` for an interactive prompt.
 
 TODO: Delete this and the text above, and describe your gem

data/bin/semtools.rb

@@ -20,6 +20,13 @@ def load_tabular_file(file)
   return records
 end
 
+def format_tabular_data(data, separator, id_col, terms_col)
+    data.map!{|row| 
+      [row[id_col], 
+       row[terms_col].split(separator).map!{|term| term.to_sym}]
+    }
+end
+
 def store_profiles(file, ontology)
   file.each do |id, terms|
     ontology.add_profile(id, terms)
@@ -85,22 +92,14 @@ def clean_profiles(profiles, ontology, options)
 	return removed_profiles
 end
 
-def expand_profiles(profiles, ontology, unwanted_terms = [])
-	profiles.each do |disease_id, terms|
-		terms.each do |term|
-	    	profiles[disease_id] << ontology.get_ancestors(term).difference(unwanted_terms)
-	  end
-	end
-end	
-
-def write_similarity_profile_list(input, onto_obj, similarity_type)
-  similarity_file = File.basename(input, ".*")+'_semantic_similarity_list'  
-  File.open(similarity_file, 'w') do |file|
-    onto_obj.profiles.each do |profile_query_key, profile_query_value|
-      onto_obj.profiles.each do |profile_search_key, profile_search_value|
-        file.puts([profile_query_key, profile_search_key, onto_obj.compare(profile_query_value, profile_search_value, sim_type: similarity_type)].join("\t"))
+def write_similarity_profile_list(output, onto_obj, similarity_type, refs)
+  profiles_similarity = onto_obj.compare_profiles(sim_type: similarity_type, external_profiles: refs)
+  File.open(output, 'w') do |f|
+    profiles_similarity.each do |pairsA, pairsB_and_values|
+      pairsB_and_values.each do |pairsB, values|
+        f.puts "#{pairsA}\t#{pairsB}\t#{values}"
       end
-    end 
+    end
   end
 end
 
@@ -156,8 +155,77 @@ def get_stats(stats)
   return report_stats
 end
 
+def sort_terms_by_levels(terms, modifiers, ontology, all_childs)
+  term_levels = ontology.get_terms_levels(all_childs)
+  if modifiers.include?('a')
+    term_levels.sort!{|t1,t2| t2[1] <=> t1[1]}
+  else
+    term_levels.sort!{|t1,t2| t1[1] <=> t2[1]}
+  end
+  all_childs = term_levels.map{|t| t.first}
+  return all_childs, term_levels
+end
 
+  def get_childs(ontology, terms, modifiers)
+    #modifiers
+    # - a: get ancestors instead of decendants
+    # - r: get parent-child relations instead of list descendants/ancestors
+    # - hN: when list of relations, it is limited to N hops from given term
+    # - n: give terms names instead of term codes
+    all_childs = []
+    terms.each do |term|
+      if modifiers.include?('a') 
+        childs = ontology.get_ancestors(term)
+      else
+        childs = ontology.get_descendants(term)
+      end
+      all_childs = all_childs | childs
+    end
+    if modifiers.include?('r')
+      relations = []
+      all_childs = all_childs | terms # Add parents that generated child list
+      target_hops = nil
+      if /h([0-9]+)/ =~ modifiers
+        target_hops = $1.to_i + 1 # take into account refernce term (parent/child) addition
+        all_childs, term_levels = sort_terms_by_levels(terms, modifiers, ontology, all_childs)
+      end
 
+      current_level = nil
+      hops = 0
+      all_childs.each_with_index do |term, i|
+        if !target_hops.nil?
+          level = term_levels[i][1]
+          if level != current_level
+            current_level = level
+            hops +=1
+            break if hops == target_hops + 1 # +1 take into account that we have detected a level change and we saved the last one entirely
+          end
+        end
+        if modifiers.include?('a')
+          descendants = ontology.get_direct_ancentors(term)
+        else
+          descendants = ontology.get_direct_descendants(term)
+        end
+        if !descendants.nil?
+          descendants.each do |desc|
+            if modifiers.include?('a')
+              relations << [desc, term]
+            else
+              relations << [term, desc]
+            end
+          end
+        end
+      end
+      all_childs = []
+      relations.each do |rel| 
+        rel, _ = ontology.translate_ids(rel) if modifiers.include?('n')
+        all_childs << rel
+      end
+    else
+      all_childs.map!{|c| ontology.translate_id(c)} if modifiers.include?('n') 
+    end
+    return all_childs
+  end
 
 
 
@@ -183,9 +251,9 @@ OptionParser.new do |opts|
     options[:output_file] = item
   end
 
-  options[:IC] = false
-  opts.on("-I", "--IC", "Get IC") do
-    options[:IC] = true
+  options[:IC] = nil
+  opts.on("-I", "--IC STRING", "Get IC. 'prof' for stored profiles or 'ont' for terms in ontology") do |item|
+    options[:IC] = item
   end
 
   options[:ontology_file] = nil
@@ -207,6 +275,11 @@ OptionParser.new do |opts|
   	options[:similarity] = sim_method.to_sym
   end
 
+  options[:reference_profiles] = nil
+  opts.on("--reference_profiles PATH", "Path to file tabulated file with first column as id profile and second column with ontology terms separated by separator. ") do |opt|
+    options[:reference_profiles] = opt
+  end
+
   options[:clean_profiles] = false
 	opts.on("-c", "--clean_profiles", "Removes ancestors, descendants and obsolete terms from profiles") do
   	options[:clean_profiles] = true
@@ -228,13 +301,13 @@ OptionParser.new do |opts|
   end
 
   options[:xref_sense] = :byValue
-  opts.on("--xref_sense ", "Ontology-xref or xref-ontology. By default xref-ontology if set, ontology-xref") do
+  opts.on("--xref_sense", "Ontology-xref or xref-ontology. By default xref-ontology if set, ontology-xref") do
     options[:xref_sense] = :byTerm
   end
 
-  options[:expand_profiles] = false
-  opts.on("-e", "--expand_profiles", "Expand profiles adding ancestors") do
-    options[:expand_profiles] = true
+  options[:expand_profiles] = nil
+  opts.on("-e", "--expand_profiles STRING", "Expand profiles adding ancestors if 'parental', adding new profiles if 'propagate'") do |meth|
+    options[:expand_profiles] = meth 
   end
 
   options[:unwanted_terms] = []
@@ -270,15 +343,19 @@ OptionParser.new do |opts|
   end     
 
   options[:subject_column] = 0
-  opts.on("-f NUM", "--subject_column NUM", "The number of the column for the subject id") do |ncol|
+  opts.on("-f NUM", "--subject_column INTEGER", "The number of the column for the subject id") do |ncol|
     options[:subject_column] = ncol.to_i
   end
 
   options[:annotations_column] = 1
-  opts.on("-a NUM", "--annotations_column NUM", "The number of the column for the annotation ids") do |ncol|
-    options[:annotations_column] = ncol.to_i
+  opts.on("-a NUM", "--annotations_column INTEGER", "The number of the column for the annotation ids") do |item|
+    options[:annotations_column] = item.to_i
   end
 
+  options[:root] = nil
+  opts.on("-R STRING", "--root STRING", "Term id to be considered the new root of the ontology") do |item|
+    options[:root] = item.to_sym
+  end
 
   options[:list_term_attributes] = false
   opts.on("--list_term_attributes", "The number of the column for the annotation ids") do
@@ -299,15 +376,17 @@ end
 if !options[:ontology_file].nil?
   options[:ontology_file] = get_ontology_file(options[:ontology_file], ont_index_file)
 end
-ontology = Ontology.new(file: options[:ontology_file], load_file: true)
+
+extra_dicts = []
+extra_dicts << [:xref, {select_regex: /(#{options[:keyword]})/, store_tag: :tag, multiterm: true}] if !options[:keyword].nil?
+ontology = Ontology.new(file: options[:ontology_file], load_file: true, extra_dicts: extra_dicts)
+
+Ontology.mutate(options[:root], ontology, clone: false) if !options[:root].nil? # TODO fix method and convert in class method
 
 if !options[:input_file].nil?
   data = load_tabular_file(options[:input_file])
   if options[:list_translate].nil? || !options[:keyword].nil?
-    data.map!{|row| 
-      [row[options[:subject_column]], 
-       row[options[:annotations_column]].split(options[:separator]).map!{|term| term.to_sym}]
-    }
+    format_tabular_data(data, options[:separator], options[:subject_column], options[:annotations_column])
     store_profiles(data, ontology) if options[:translate] != 'codes' && options[:keyword].nil?
   end
 end
@@ -345,16 +424,24 @@ if options[:clean_profiles]
 	end
 end
 
-if options[:expand_profiles]
-  expanded_profiles = expand_profiles(ontology.profiles, ontology, options[:unwanted_terms])
+if !options[:expand_profiles].nil?
+  ontology.expand_profiles(options[:expand_profiles], unwanted_terms: options[:unwanted_terms])
 end 
 
 if !options[:similarity].nil?
-  write_similarity_profile_list(input = options[:input_file], onto_obj=ontology, similarity_type = options[:similarity])
+  refs = nil
+  if !options[:reference_profiles].nil?
+    refs = load_tabular_file(options[:reference_profiles])
+    format_tabular_data(refs, options[:separator], 0, 1)
+    refs = refs.to_h
+    refs = clean_profiles(ontology.profiles, ontology, options) if options[:clean_profiles]
+    abort('Reference profiles are empty after cleaning ') if refs.nil? || refs.empty?
+  end
+  write_similarity_profile_list(options[:output_file], ontology, options[:similarity], refs)
 end 
 
 
-if options[:IC]
+if options[:IC] == 'prof'
   ontology.add_observed_terms_from_profiles
   by_ontology, by_freq = ontology.get_profiles_resnik_dual_ICs
   ic_file = File.basename(options[:input_file], ".*")+'_IC_onto_freq'
@@ -363,6 +450,12 @@ if options[:IC]
         file.puts([id, by_ontology[id], by_freq[id]].join("\t"))
     end       
   end
+elsif options[:IC] == 'ont'
+  File.open('ont_IC' , 'w') do |file|
+    ontology.each do |term|
+        file.puts "#{term}\t#{ontology.get_IC(term)}"
+    end       
+  end
 end    
 
 if options[:translate] == 'names'
@@ -371,38 +464,17 @@ end
 
 if !options[:childs].first.empty?
   terms, modifiers = options[:childs]
-  all_childs = []
-  terms.each do |term|
-   childs = ontology.get_descendants(term)
-   all_childs = all_childs | childs
-  end
-  if modifiers.include?('r')
-    relations = []
-    all_childs = all_childs | terms # Add parents that generated child list
-    all_childs.each do |term|
-      descendants = ontology.get_direct_descendants(term)
-      if !descendants.nil?
-        descendants.each do |desc|
-          relations << [term, desc]
-        end
-      end
-    end
-    relations.each do |rel| 
-      rel, _ = ontology.translate_ids(rel) if modifiers.include?('n')
-      puts rel.join("\t")
-    end
-  else
-    all_childs.each do |c| 
-      if modifiers.include?('n') 
-        puts ontology.translate_id(c) 
-      else
-        puts c
-      end
+  all_childs = get_childs(ontology, terms, modifiers)
+  all_childs.each do |ac|
+    if modifiers.include?('r')
+      puts ac.join("\t")
+    else
+      puts ac
     end
   end
 end
 
-if !options[:output_file].nil?
+if !options[:output_file].nil? && options[:similarity].nil?
   File.open(options[:output_file], 'w') do |file|
     ontology.profiles.each do |id, terms|
       file.puts([id, terms.join("|")].join("\t"))
@@ -426,12 +498,15 @@ end
 
 if !options[:keyword].nil?
   xref_translated = []
-  ontology.calc_dictionary(:xref, select_regex: /(#{options[:keyword]})/, store_tag: :tag, multiterm: true, substitute_alternatives: false)
   dict = ontology.dicts[:tag][options[:xref_sense]]
   data.each do |id, prof|
     xrefs = []
     prof.each do |t|
-      query = dict[t.to_s]
+	if options[:xref_sense] == :byValue
+	      query = dict[t.to_s]
+	else
+	      query = dict[t]
+	end
       xrefs.concat(query) if !query.nil?
     end
     xref_translated << [id, xrefs] if !xrefs.empty?
@@ -443,4 +518,4 @@ if !options[:keyword].nil?
       end
     end
   end
-end
+end