version 1.1.1

build/lib/paste/PASTE.py

@@ -3,8 +3,9 @@
 import ot
 from sklearn.decomposition import NMF
 from scipy.spatial import distance_matrix
+import scipy
 from numpy import linalg as LA
-from .helper import kl_divergence, intersect
+from .helper import kl_divergence, intersect, to_dense_array
 
 def pairwise_align(sliceA, sliceB, alpha = 0.1, G_init = None, a_distribution = None, b_distribution = None, norm = False, numItermax = 200, return_obj = False, verbose = False, **kwargs):
     """
@@ -33,8 +34,8 @@ def pairwise_align(sliceA, sliceB, alpha = 0.1, G_init = None, a_distribution =
 
     D_A = distance_matrix(sliceA.obsm['spatial'], sliceA.obsm['spatial'])
     D_B = distance_matrix(sliceB.obsm['spatial'], sliceB.obsm['spatial'])
-    s_A = sliceA.X + 0.01
-    s_B = sliceB.X + 0.01
+    s_A = to_dense_array(sliceA.X) + 0.01
+    s_B = to_dense_array(sliceB.X) + 0.01
     M = kl_divergence(s_A, s_B)
 
     if a_distribution is None:
@@ -48,8 +49,8 @@ def pairwise_align(sliceA, sliceB, alpha = 0.1, G_init = None, a_distribution =
         b = b_distribution
 
     if norm:
-        D1 /= D1[D1>0].min().min()
-        D2 /= D2[D2>0].min().min()
+        D_A /= D_A[D_A>0].min().min()
+        D_B /= D_B[D_B>0].min().min()
 
     if G_init is None:
         pi, logw = ot.gromov.fused_gromov_wasserstein(M, D_A, D_B, a, b, loss_fun='square_loss', alpha= alpha, log=True, numItermax=numItermax,verbose=verbose)
@@ -99,7 +100,7 @@ def center_align(A, slices, lmbda, alpha = 0.1, n_components = 15, threshold = 0
         W = model.fit_transform(A.X)
     else:
         pis = pis_init
-        W = model.fit_transform(A.shape[0]*sum([lmbda[i]*np.dot(pis[i], slices[i].X) for i in range(len(slices))]))
+        W = model.fit_transform(A.shape[0]*sum([lmbda[i]*np.dot(pis[i], to_dense_array(slices[i].X)) for i in range(len(slices))]))
     H = model.components_
     center_coordinates = A.obsm['spatial']
 
@@ -146,7 +147,7 @@ def center_ot(W, H, slices, center_coordinates, common_genes, alpha, norm = Fals
 def center_NMF(W, H, slices, pis, lmbda, n_components, random_seed, verbose = False):
     print('Solving Center Mapping NMF Problem:')
     n = W.shape[0]
-    B = n*sum([lmbda[i]*np.dot(pis[i], slices[i].X) for i in range(len(slices))])
+    B = n*sum([lmbda[i]*np.dot(pis[i], to_dense_array(slices[i].X)) for i in range(len(slices))])
     model = NMF(n_components=n_components, solver = 'mu', beta_loss = 'kullback-leibler', init='random', random_state = random_seed, verbose = verbose)
     W_new = model.fit_transform(B)
     H_new = model.components_

build/lib/paste/helper.py

@@ -35,20 +35,17 @@ def kl_divergence(X, Y):
     return np.asarray(D)
 
 
-def intersect(a, b): 
+def intersect(lst1, lst2): 
     """
-    param: a - list
-    param: b - list
+    param: lst1 - list
+    param: lst2 - list
     
     return: list of common elements
     """
-    a_set = set(a) 
-    b_set = set(b) 
-    if (a_set & b_set): 
-        return list(a_set & b_set) 
-    else: 
-        print("No common elements")
-
+
+    temp = set(lst2)
+    lst3 = [value for value in lst1 if value in temp]
+    return lst3 
 
 def norm_and_center_coordinates(X): 
     """
@@ -77,4 +74,6 @@ def match_spots_using_spatial_heuristic(X,Y,use_ot=True):
         pi[row_ind, col_ind] = 1/max(n1,n2)
         if n1<n2: pi[:, [(j not in col_ind) for j in range(n2)]] = 1/(n1*n2)
         elif n2<n1: pi[[(i not in row_ind) for i in range(n1)], :] = 1/(n1*n2)
-    return pi
+    return pi
+
+to_dense_array = lambda X: np.array(X.todense()) if isinstance(X,scipy.sparse.csr.spmatrix) else X

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = paste-bio
-version = 1.1.0
+version = 1.1.1
 author = Max Land
 author_email = max.ruikang.land@gmail.com
 description = A computational method to align and integrate spatial transcriptomics experiments.

src/paste_bio.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: paste-bio
-Version: 1.1.0
+Version: 1.1.1
 Summary: A computational method to align and integrate spatial transcriptomics experiments.
 Home-page: https://github.com/raphael-group/paste
 Author: Max Land
@@ -17,7 +17,9 @@ License-File: LICENSE
 
 # PASTE
 
-PASTE is a computational method that leverages both gene expression similarity and spatial distances between spots align and integrate spatial transcriptomics data. In particular, there are two methods:
+![PASTE Overview](https://github.com/raphael-group/paste/blob/main/paste_overview.png)
+
+PASTE is a computational method that leverages both gene expression similarity and spatial distances between spots to align and integrate spatial transcriptomics data. In particular, there are two methods:
 1. `pairwise_align`: align spots across pairwise slices.
 2. `center_align`: integrate multiple slices into one center slice.
 
@@ -29,6 +31,8 @@ PASTE is actively being worked on with future updates coming.
 
 As of version 1.1.0, PASTE now runs on AnnData making it very easy to integrate with Scanpy for better downstream analysis. Hooray!
 
+This also means that the old version that uses the `STLayer` object is now deprecated.
+
 ### Dependencies
 
 To run PASTE, you will need the following Python packages:
@@ -41,13 +45,17 @@ To run PASTE, you will need the following Python packages:
 
 ### Installation
 
-The easiest way is to install PASTE on pypi: https://pypi.org/project/paste-bio/.
+The easiest way is to install PASTE on pypi: https://pypi.org/project/paste-bio/. 
+
+`pip install paste-bio` 
+
+Or you can install PASTE on bioconda: https://anaconda.org/bioconda/paste-bio.
 
-`pip install paste-bio`
+`conda install -c bioconda paste-bio`
 
 Check out Tutorial.ipynb for an example of how to use PASTE.
 
-Or you can clone the respository and run from command line (see below).
+Lastly, you can clone the respository and run from command line (see below).
 
 
 ### Command Line