version 1.0

docs/release_notes.rst

@@ -10,24 +10,24 @@ See the documentation of version 0.4.4 `here <http://scanpy.readthedocs.io/en/0.
 
 **March 28, 2018**: version 1.0
 
-Scanpy is much faster. A standard analysis of 130k cells now takes about `14 min
+Scanpy is much faster and memory efficient. A clustering analysis of 130k cells now takes about `14 min
 <https://github.com/theislab/scanpy_usage/blob/master/170522_visualizing_one_million_cells/logfile_130k.txt>`_. A
-standard analysis of 1.3M cells takes about `6 h
+clustering analysis of 1.3M cells takes about `6 h
 <https://github.com/theislab/scanpy_usage/blob/master/170522_visualizing_one_million_cells/logfile_1.3M.txt>`_.
 
 The API gained a preprocessing function :func:`~scanpy.api.pp.neighbors` and a
 class :func:`~scanpy.api.Neighbors` to which all basic graph computations are
-delegated. By that, the used data representation in pipelines is more
-transparent and tools are less bloated with parameters.
+delegated. By that, the used data representation in pipelines became more
+transparent and tools less bloated with parameters.
 
 .. warning::
 
    Upgrading to 1.0 isn't fully backwards compatible.
 
    - the graph-based tools :func:`~scanpy.api.tl.louvain`
-   :func:`~scanpy.api.tl.dpt` :func:`~scanpy.api.tl.draw_graph`
-   :func:`~scanpy.api.tl.umap` :func:`~scanpy.api.tl.diffmap`
-   :func:`~scanpy.api.tl.paga` now require a prior computtion of the graph::
+     :func:`~scanpy.api.tl.dpt` :func:`~scanpy.api.tl.draw_graph`
+     :func:`~scanpy.api.tl.umap` :func:`~scanpy.api.tl.diffmap`
+     :func:`~scanpy.api.tl.paga` now require prior computation of the graph::
      
          sc.pp.neighbors(adata, n_neighbors=5)
          sc.tl.louvain(adata)
@@ -57,17 +57,18 @@ transparent and tools are less bloated with parameters.
    - upgrading scikit-learn from 0.18 to 0.19 changed the implementation of PCA,
      some results might therefore look slightly different
 
-Further changes are as follows
+Further changes are
 
-- UMAP [McInnes18]_ can serve as a first visualization of the data just as
-  tSNE. In contrast to tSNE, UMAP directly embeds the single-cell graph and is
-  faster.
+- UMAP [McInnes18]_ can serve as a first visualization of the data just as tSNE,
+  in contrast to tSNE, UMAP directly embeds the single-cell graph and is faster;
+  UMAP is now also used for measuring connectivities and computing neighbors,
+  see :func:`~scanpy.api.pp.neighbors`
 
-- graph abstraction: AGA is renamed to PAGA :func:`~scanpy.api.tl.paga`. now, it
-  only measures connectivities between partitions of the single-cell graph,
+- graph abstraction: AGA is renamed to PAGA: :func:`~scanpy.api.tl.paga`; now,
+  it only measures connectivities between partitions of the single-cell graph,
   pseudotime and clustering need to be computed separately via
   :func:`~scanpy.api.tl.louvain` and :func:`~scanpy.api.tl.dpt`, the
-  connectivity measure has been improved  
+  connectivity measure has been improved
 
 - logistic regression for finding marker genes
   :func:`~scanpy.api.tl.rank_genes_groups` with parameter `metfod='logreg'`

scanpy/api/__init__.py

@@ -37,8 +37,6 @@
 
 **Neighbors**
 
-Compute a neighborhood-graph representation of the data, usually a knn-graph.
-
 .. autosummary::
    :toctree: .
 
@@ -59,7 +57,7 @@
    tl.draw_graph
    tl.diffmap
 
-**Clustering, branching trajectories and pseudotime based on single-cell neighborhood graph**
+**Clustering, branching trajectories and pseudotime based on single-cell graph**
 
 .. autosummary::
    :toctree: .
@@ -134,14 +132,14 @@
 .. autosummary::
    :toctree: .
 
-   pl.aga
-   pl.aga_graph
-   pl.aga_path
    pl.louvain
    pl.dpt
    pl.dpt_scatter
    pl.dpt_groups_pseudotime
    pl.dpt_timeseries
+   pl.paga
+   pl.paga_path
+   pl.paga_compare
    pl.rank_genes_groups
    pl.rank_genes_groups_violin
 
@@ -233,9 +231,7 @@
 Settings and Logging
 --------------------
 
-Global settings.
-
-One function `settings.set_figure_params`.
+Note the functions `settings.set_figure_params` and `logging.print_versions()`.
 
 ==============================================  ===================================
 `settings.verbosity`                            Verbosity level (default: 1).
@@ -256,13 +252,6 @@
 ...  Show even more detailed progress.
 ===  ======================================
 
-Logging.
-
-================================================  ===================================
-`logging.print_version_and_date()`                Print the version and the date.
-`logging.print_versions_dependencies_numerics()`  Print the versions of dependencies.
-================================================  ===================================
-
 """
 
 from anndata import AnnData

scanpy/neighbors/__init__.py

@@ -26,7 +26,7 @@ def neighbors(
         metric_kwds={},
         copy=False):
     """\
-    Compute a neighborhood graph of data points [McInnes18]_.
+    Compute a neighborhood graph of observations [McInnes18]_.
 
     The neighbor search efficiency of this heavily relies on UMAP [McInnes18]_,
     which also provides a method for estimating connectivities of data points -

scanpy/tools/paga.py

@@ -8,10 +8,9 @@
 from .. import settings
 
 
-def paga(adata,
+def paga(
+        adata,
         groups='louvain',
-        # threshold=0.01,
-        tree_based_confidence=False,
         n_jobs=None,
         copy=False):
     """\
@@ -33,12 +32,6 @@ def paga(adata,
         graph. 'louvain_groups' uses the Louvain algorithm and optimizes
         modularity of the graph. You can also pass your predefined groups by
         choosing any categorical annotation of observations (`adata.obs`).
-    tree_based_confidence : `bool`, optional (default: `True`)
-        Have high confidence in a connection if its connectivities is
-        significantly higher than the median connectivities of the global spanning
-        tree of the abstracted graph.
-    n_jobs : `int` or None (default: `sc.settings.n_jobs`)
-        Number of CPUs to use for parallel processing.
     copy : `bool`, optional (default: `False`)
         Copy `adata` before computation and return a copy. Otherwise, perform
         computation inplace and return None.
@@ -62,7 +55,7 @@ def paga(adata,
     adata = adata.copy() if copy else adata
     utils.sanitize_anndata(adata)
     logg.info('running partition-based graph abstraction (PAGA)', reset=True)
-    paga = PAGA(adata, groups, tree_based_confidence=tree_based_confidence)
+    paga = PAGA(adata, groups)
     paga.compute()
     adata.uns['paga'] = {}
     adata.uns['paga']['connectivities'] = paga.connectivities_coarse
@@ -80,10 +73,8 @@ def paga(adata,
 
 
 class PAGA(Neighbors):
-    """Statistical Graph Abstraction.
-    """
 
-    def __init__(self, adata, groups, threshold=0.05, tree_based_confidence=True):
+    def __init__(self, adata, groups, threshold=0.05, tree_based_confidence=False):
         super(PAGA, self).__init__(adata)
         self._groups = groups
         self.threshold = threshold