Make dependencies hard, clean up install docs

docs/install.rst

@@ -8,8 +8,6 @@ For the impatient
 
     virtualenv kevlar-env
     source kevlar-env/bin/activate  # Execute this with every new terminal session
-    # pip install pysam  # You probably don't need this
-    pip install networkx
     pip install git+https://github.com/dib-lab/khmer.git
     pip install biokevlar
 
@@ -25,27 +23,29 @@ The command :code:`source kevlar-env/bin/activate` will need to be re-executed a
 Dependencies
 ------------
 
-Upcoming releases of **kevlar** will install dependencies automatically, but for now these must be installed manually.
-Currently the only hard non-standard dependency is the **khmer** library—to be precise, its :code:`feature/consume_bitsplit` development branch.
-The :code:`kevlar dump` command also depends on **pysam**, while :code:`kevlar filter` and :code:`kevlar assemble` depend on **networkx**.
+The **kevlar** software has three non-standard dependencies: the `networkx package <https://networkx.github.io/>`_, the `pysam <http://pysam.readthedocs.io/>` package, and the `khmer package <http://khmer.readthedocs.io/>`.
+Currently, the **kevlar** installation procedure handles the first two dependencies automatically, but since it relies on an unreleased version of **khmer** this last dependency must be install manually.
 
 .. code::
 
-    pip install pysam
-    pip install networkx
     pip install git+https://github.com/dib-lab/khmer.git
 
-The **pysam** dependency will eventually be dropped.
-
 Installation
 ------------
 
-Once the dependencies are installed, **kevlar** can be installed with the :code:`pip` command.
+Once **khmer** is installed, **kevlar** can be installed with the :code:`pip` command.
 
 .. code::
 
     pip install biokevlar
 
+This installs the most recent stable release.
+If you want to install the latest (possibly unstable) version, pip can install **kevlar** directly from GitHub.
+
+.. code:
+
+    pip install git+https://github.com/dib-lab/kevlar.git
+
 If you want to test whether kevlar is installed and running correctly, use pytest.
 
 .. code::

kevlar/assemble.py

@@ -11,10 +11,7 @@
 from collections import defaultdict, namedtuple
 import itertools
 import sys
-try:
-    import networkx
-except:
-    pass
+import networkx
 import screed
 import khmer
 import kevlar

kevlar/cli/filter.py

@@ -90,10 +90,6 @@ def subparser(subparsers):
                            help='output file; default is terminal (stdout)')
     misc_args.add_argument('--aug-out', metavar='FILE',
                            help='optional augmented Fastq output')
-    misc_args.add_argument('--cc-prefix', metavar='PREFIX',
-                           help='group reads by novel k-mers, and use the '
-                           'specified prefix to write each group to its own '
-                           'file')
 
     subparser.add_argument('augfastq', nargs='+', help='one or more files in '
                            '"augmented" Fastq format (a la `kevlar novel` '

kevlar/dump.py

@@ -8,10 +8,7 @@
 # -----------------------------------------------------------------------------
 
 from __future__ import print_function
-try:
-    import pysam
-except ImportError:
-    pass
+import pysam
 import khmer
 from khmer import khmer_args
 
@@ -24,14 +21,6 @@ def match(record, seq):
 
 
 def main(args):
-    try:  # pragma: no cover
-        import pysam
-    except ImportError:
-        import sys
-        print('[kevlar::dump] FATAL ERROR: cannot execute "kevlar dump" '
-              'unless the "pysam" module is installed', file=sys.stderr)
-        sys.exit(1)
-
     print('[kevlar::dump] Loading reference sequence', file=args.logfile)
     with kevlar.open(args.refr, 'r') as genome:
         seqs = kevlar.seqio.parse_seq_dict(genome)

kevlar/filter.py

@@ -117,15 +117,6 @@ def validate_and_print(readset, countgraph, refr=None, contam=None, minabund=5,
 
 
 def main(args):
-    if args.cc_prefix:  # pragma: no cover
-        try:
-            import networkx
-        except ImportError:
-            print('[kevlar::filter] FATAL ERROR: cannot group reads by novel '
-                  'k-mers (--cc-prefix flag) unless the "networkx" module is '
-                  'installed', file=sys.stderr)
-            sys.exit(1)
-
     timer = kevlar.Timer()
     timer.start()
 
@@ -175,16 +166,6 @@ def main(args):
     print('[kevlar::filter] k-mers validated and reads printed',
           'in {:.2f} sec'.format(elapsed), file=args.logfile)
 
-    if args.cc_prefix:
-        timer.start('graph')
-        print('[kevlar::filter] Group reads by novel k-mers',
-              file=args.logfile)
-        readset.group_reads_by_novel_kmers(args.cc_prefix,
-                                           logstream=args.logfile)
-        elapsed = timer.stop('graph')
-        print('[kevlar::filter] reads grouped by novel k-mers',
-              'in {:.2f} sec'.format(elapsed), file=args.logfile)
-
     total = timer.stop()
     message = 'Total time: {:.2f} seconds'.format(total)
     print('[kevlar::filter]', message, file=args.logfile)

kevlar/overlap.py

@@ -12,10 +12,7 @@
 from collections import defaultdict, namedtuple
 import itertools
 import sys
-try:
-    import networkx
-except:
-    pass
+import networkx
 import screed
 import kevlar
 

kevlar/seqio.py

@@ -10,10 +10,7 @@
 from __future__ import print_function
 from collections import defaultdict
 from itertools import combinations, product
-try:
-    from networkx import Graph, connected_components
-except:
-    pass
+from networkx import Graph, connected_components
 from sys import stdout, stderr, exit
 import re
 import khmer
@@ -214,15 +211,6 @@ def recalc_abund(self, newcounts, minabund=5):
 
     def group_reads_by_novel_kmers(self, ccprefix, upint=10000,
                                    logstream=None):
-        try:
-            _ = Graph()
-        except NameError as ne:
-            assert "'Graph' is not defined" in str(ne)
-            print('[kevlar::AnnotatedReadSet] FATAL ERROR: Cannot call '
-                  '"group_reads_by_novel_kmers" unless the "networkx" module '
-                  ' is installed', file=stderr)
-            exit(1)
-
         n = 0
         reads_by_novel_kmer = defaultdict(set)
         novel_kmers = set()  # just for reporting numbers; free memory ASAP

setup.py

@@ -25,6 +25,8 @@
           'kevlar': ['kevlar/tests/data/*', 'kevlar/tests/data/*/*']
       },
       include_package_data=True,
+      setup_requires=['pysam', 'networkx'],
+      install_requires=['pysam', 'networkx'],
       entry_points={
           'console_scripts': ['kevlar = kevlar.__main__:main']
       },