Merge pull request #7 from apcamargo/individualized-tsi

Version 0.5.0

README.md

@@ -50,9 +50,9 @@ positional arguments:
   input_file            Expression matrix file in the TSV or CSV formats.
   output_file           Output TSV file containing tissue-specificity values.
   method                Tissue-specificity metric. Allowed values are:
-                        "counts", "tsi", "tau", "gini", "simpson",
-                        "shannon_specificity", "roku_specificity", "zscore",
-                        "spm", "spm_dpm", "js_specificity",
+                        "counts", "tau", "gini", "simpson",
+                        "shannon_specificity", "roku_specificity", "tsi",
+                        "zscore", "spm", "spm_dpm", "js_specificity",
                         "js_specificity_dpm".
 
 optional arguments:

setup.py

@@ -25,7 +25,7 @@
 
 setup(
     name='tspex',
-    version='0.4.0',
+    version='0.5.0',
     packages=find_packages(),
     license='GNU General Public License v3.0',
     description='A Python package for calculating tissue-specificity metrics for gene expression.',

tspex/cli.py

@@ -39,9 +39,8 @@ def tspex_cli(input_file, output_file, method, log, disable_transformation, thre
 
 
 def main():
-    method_choices = ['counts', 'tsi', 'tau', 'gini', 'simpson', 'shannon_specificity',
-                      'roku_specificity', 'zscore', 'spm', 'spm_dpm', 'js_specificity',
-                      'js_specificity_dpm']
+    method_choices = ['counts', 'tau', 'gini', 'simpson', 'shannon_specificity', 'roku_specificity',
+                      'tsi', 'zscore', 'spm', 'spm_dpm', 'js_specificity', 'js_specificity_dpm']
     parser = argparse.ArgumentParser(
         description='Compute gene tissue-specificity from an expression matrix and save the output.',
         formatter_class=argparse.ArgumentDefaultsHelpFormatter)

tspex/core/specificity_class.py

@@ -46,9 +46,9 @@ class TissueSpecificity:
         corresponding to genes and columns to tissues/conditions.
     method : str
         A string representing which tissue-expression metric should be
-        calculated. One of: 'counts', 'tsi', 'tau', 'gini', 'simpson',
-        'shannon_specificity', 'roku_specificity', 'zscore', 'spm', 'spm_dpm',
-        'js_specificity', 'js_specificity_dpm'.
+        calculated. One of: 'counts', 'tau', 'gini', 'simpson',
+        'shannon_specificity', 'roku_specificity', 'tsi', 'zscore', 'spm',
+        'spm_dpm', 'js_specificity', 'js_specificity_dpm'.
     log : bool, default False
         Log-transform the expression matrix before computing tissue-specificity
         by taking the base-2 logarithm of one plus the expression values. By
@@ -76,12 +76,12 @@ class TissueSpecificity:
     def __init__(self, expression_data, method, log=False, **kwargs):
         self._function_dictionary = {
             'counts': counts,
-            'tsi': tsi,
             'tau': tau,
             'gini': gini,
             'simpson': simpson,
             'shannon_specificity': shannon_specificity,
             'roku_specificity': roku_specificity,
+            'tsi': tsi,
             'zscore': zscore,
             'spm': spm,
             'spm_dpm': spm_dpm,
@@ -105,7 +105,7 @@ def __init__(self, expression_data, method, log=False, **kwargs):
 
     def _compute_tissue_specificity(self):
         func = self._function_dictionary[self._method]
-        if self._method in ['zscore', 'spm', 'js_specificity']:
+        if self._method in ['tsi', 'zscore', 'spm', 'js_specificity']:
             tissue_specificity = self.expression_data.apply(func, axis=1, result_type='broadcast',
                                                             transform=self._transform)
         else:
@@ -118,8 +118,8 @@ def _compute_tissue_specificity(self):
     def plot_histogram(self, bins=30, size=(6, 4), dpi=75):
         """
         Plot a histogram of the tissue-specificity values. If the chosen metric
-        is one of 'zscore', 'spm' or 'js_specificity', the maximum row value is used
-        as a representative of the gene tissue-specificity.
+        is one of 'tsi', 'zscore', 'spm' or 'js_specificity', the maximum row
+        value is used as a representative of the gene tissue-specificity.
 
         Parameters
         ----------
@@ -132,7 +132,7 @@ def plot_histogram(self, bins=30, size=(6, 4), dpi=75):
         """
 
         with plt.style.context('seaborn-whitegrid'):
-            if self._method in ['zscore', 'spm', 'js_specificity']:
+            if self._method in ['tsi', 'zscore', 'spm', 'js_specificity']:
                 data = self.tissue_specificity.max(axis=1).values
             else:
                 data = self.tissue_specificity.values
@@ -147,8 +147,9 @@ def plot_heatmap(self, threshold, sort_genes=False, use_zscore=False, gene_names
         """
         Plot a heatmap of the expression of genes with tissue-specificity over a
         given a threshold. The threshold should be in the [0,1] range. If the
-        chosen metric is one of 'zscore', 'spm' or 'js_specificity', the maximum
-        row value is used as a representative of the gene tissue-specificity.
+        chosen metric is one of 'tsi', 'zscore', 'spm' or 'js_specificity', the
+        maximum row value is used as a representative of the gene
+        tissue-specificity.
 
         Parameters
         ----------
@@ -170,7 +171,7 @@ def plot_heatmap(self, threshold, sort_genes=False, use_zscore=False, gene_names
             The resolution in dots per inch.
         """
 
-        if self._method in ['zscore', 'spm', 'js_specificity']:
+        if self._method in ['tsi', 'zscore', 'spm', 'js_specificity']:
             ts_data = self.tissue_specificity.max(axis=1)
         else:
             ts_data = self.tissue_specificity

tspex/core/specificity_functions.py

@@ -82,36 +82,6 @@ def counts(vector, **kwargs):
             return cts_transformed
 
 
-def tsi(vector, **kwargs):
-    """
-    Quantify tissue-specificity as the ratio between the vector's maximum
-    expression value and the sum of the expression values in all tissues.
-
-    Parameters
-    ----------
-    vector : numpy.array
-        Gene expression vector. Each value corresponds to the gene expression
-        in a given tissue.
-
-    Returns
-    -------
-    float
-        Single summary of the tissue-specificity. Ranges from 0 (ubiquitous
-        expression) to 1 (specific expression).
-
-    References
-    ----------
-    .. [1] Julien, Philippe, et al. "Mechanisms and evolutionary patterns of
-           mammalian and avian dosage compensation." PLoS biology 10.5 (2012)
-    """
-
-    if not np.any(vector):
-        return 0.0
-    else:
-        tissue_specificity_index = max(vector) / np.sum(vector)
-        return tissue_specificity_index
-
-
 def tau(vector, **kwargs):
     """
     Quantify tissue-specificity as the Tau index [1].
@@ -307,6 +277,36 @@ def roku_specificity(vector, **kwargs):
             return rs
 
 
+def tsi(vector, **kwargs):
+    """
+    Quantify tissue-specificity as the ratio between the expression vector
+    and the sum of the expression values in all tissues.
+
+    Parameters
+    ----------
+    vector : numpy.array
+        Gene expression vector. Each value corresponds to the gene expression
+        in a given tissue.
+
+    Returns
+    -------
+    numpy.array
+        Summary of the tissue-specificity level in each tissue. It ranges from
+        0 (ubiquitous expression) to 1 (specific expression).
+
+    References
+    ----------
+    .. [1] Julien, Philippe, et al. "Mechanisms and evolutionary patterns of
+           mammalian and avian dosage compensation." PLoS biology 10.5 (2012)
+    """
+
+    if not np.any(vector):
+        return 0.0
+    else:
+        tissue_specificity_index = vector / np.sum(vector)
+        return tissue_specificity_index
+
+
 def zscore(vector, **kwargs):
     """
     Quantify tissue-specificity as z-scores.