category data and plot selection update|

system/dataset.py

@@ -1,5 +1,5 @@
 import logger
-from scipy import array,take,asarray,shape
+from scipy import array,take,asarray,shape,nonzero
 import project
 from itertools import izip 
 
@@ -10,19 +10,14 @@ class Dataset:
     A Dataset is an n-way array with defined string identifiers across
     all dimensions.
     """
-    def __init__(self,input_array,def_list,parents=None):
+    def __init__(self,input_array,def_list):
         self._data = asarray(input_array)
         self.dims = shape(self._data)
-        self.parents = parents
         self.def_list = def_list
         self._ids_set = set()
         self.ids={}
-        self.children=[]
         self._dim_num = {}
         self._dim_names = []
-        if parents!=None:
-            for parent in self.parents:
-                parent.children.append(self)
         if len(def_list)!=len(self.dims):
             raise ValueError,"array dims and identifyer mismatch"
         for axis,(dim_name,ids) in enumerate(def_list):
@@ -37,11 +32,6 @@ class Dataset:
             self._ids_set = self._ids_set.union(set(ids))
             self._dim_num[dim_name] = axis
             self._dim_names.append(dim_name)
-            #if dim_name in project.c_p.dim_names:
-            #    if ids:
-            #        # check that identifers are same as before
-            #        raise NotImplementedError
-            #    else:
                     
         for df,d in izip(def_list,self.dims): #check that data and labels match 
             df=df[1]
@@ -75,10 +65,50 @@ class Dataset:
         n_dim = self.dims[axis]
         return [str(axis) + '_' + str(i) for i in range(n_dim)]
 
-    def index_to_id(self,dim_name,index):
+    def extract_id_from_index(self,dim_name,index):
+        """Returns a set of ids from array/list of indexes."""
         dim_ids = self.ids[dim_name]
-        return [id for id,ind in dim_ids.items() if ind in index]
+        return set([id for id,ind in dim_ids.items() if ind in index])
+
+    def extract_index_from_id(self,dim_name,id):
+        """Returns an array of indexes from a set/list of identifiers
+        (or a single id)"""
+        dim_ids = self.ids[dim_name]
+        return array([ind for name,ind in dim_ids.items() if name in id])
+        
     
+class CategoryDataset(Dataset):
+    def __init__(self,array,def_list):
+        Dataset.__init__(self,array,def_list)
+
+    def get_elements_by_category(self,dim,category):
+        """Returns all elements along input dim belonging to category.
+        Assumes a two-dim category data only!
+        """
+        if type(category)!=list:
+            raise ValueError, "category must be list"
+        gene_ids = []
+        axis_dim = self._dim_num[dim]
+        cat_index = self.extract_index_from_id(category)
+        for ind in cat_index:
+            if axis_dim==0:
+                gene_indx = nonzero(self._data[:,ind])
+            elif axis_dim==1:
+                gene_indx = nonzero(self._data[ind,:])
+            else:
+                ValueError, "Only support for 2-dim data"
+            gene_ids.append(self.extract_id_from_index(dim,gene_index))
+        return gene_ids
+                
+        
+        
+                
+            
+        
+        
+
+    
+
 class Selection:
     """Handles selected identifiers along each dimension of a dataset"""
     def __init__(self):