clean up

INSTALL

@@ -19,5 +19,22 @@ Once you have a venv-physcraper directory, you can run:
     source venv-physcraper/bin/activate
     ## This force reinstalling numpy first is key
     pip install --force-reinstall numpy==1.14.5
-    pip install -r requiremenets.txt
+    cd physcraper
+    pip install -r requirements.txt
     python setup.py install
+
+Dependencies:
+
+    Currently complete phylogenetic updating requires
+    raxmlHPC and muscle to be installed and in the path.
+
+    Check if they are installed using 
+
+    which muscle
+    which raxmlHPC
+
+    You can generate updated alignments to analyze using other phylogenetic software by creating a Scrape object and 
+    running scrape.align__query_seqs()
+
+
+

docs/examples/data_scrape.py

@@ -82,5 +82,4 @@
 
 #scraper.read_blast_wrapper()
 scraper.est_full_tree()
-
-scraper.data.write_labelled
+scraper.data.write_labelled()

physcraper/opentree_helpers.py

@@ -59,10 +59,13 @@ def bulk_tnrs_load(filename, ids_obj = None):
         for source in name.get("taxonomicSources"):
             taxsrc = source.split(":")
             otu_dict[otu]["^{}:taxon".format(taxsrc[0])] = source.strip(taxsrc[1])
+    for otu in otu_dict:
+        otu_dict[otu]["^physcraper:status"] = "original"
+        otu_dict[otu]["^physcraper:last_blasted"] = None
     return otu_dict
 
 
-def get_tree_from_synth(ott_ids, label_format="name", citation="cites.txt",):
+def get_tree_from_synth(ott_ids, label_format="name", citation="cites.txt"):
     assert label_format in ['id', 'name', 'name_and_id']
     resp = treemachine.get_synth_tree_pruned(ott_ids=ott_ids, label_format=label_format)
     cites = ''
@@ -85,6 +88,17 @@ def get_tree_from_synth(ott_ids, label_format="name", citation="cites.txt",):
     return tre
 
 
+def get_tree_from_study(study_id, tree_id, label_format="name", citation="cites.txt"):
+    assert label_format in ['id', 'name', 'name_and_id']
+    api_wrapper = APIWrapper()
+    resp = api_wrapper.study.get(study_id, tree=tree_id, format='newick', exact=True)
+    query = {"ot:studyId":study_id}
+    new_cite = oti.find_studies(query_dict = query, verbose=True)
+    cites = to_string(new_cite[0]['ot:studyPublicationReference']) + new_cite[0]['ot:studyPublication']
+    tre = Tree.get(data=resp,
+                   schema="newick")
+    return tre, cites
+
 
 
 # ATT is a dumb acronym for Alignment Tree Taxa object

physcraper/scrape.py

@@ -284,7 +284,7 @@ def run_blast_wrapper(self):  # TODO Should this be happening elsewhere?
 #            tmpfile.write(seq.symbols_as_string())
 #            tmpfile.write("\n")
             assert otu_id in self.data.otu_dict
-            last_blast = self.data.otu_dict[otu_id]['^physcraper:last_blasted']
+            last_blast = self.data.otu_dict[otu_id].get('^physcraper:last_blasted')
             if last_blast == None:
                 time_passed = delay + 1
             else:

tests/test_treetaxon.py

@@ -22,10 +22,14 @@ def test_tree_taxon():
     assert filecmp.cmp(treeout, expected_tree)
     os.remove(treeout)
 
-def test_get_tree():
+def test_get_synth_tree():
     ott_ids = set([exp_otu_dict[otu].get("^ot:ottId") for otu in exp_otu_dict])
     ott_ids = list(ott_ids)
     resp = opentree_helpers.get_tree_from_synth(ott_ids=ott_ids)
     print resp
 
-test_get_tree()
+
+def test_get_phyle_tree():
+    tr = opentree_helpers.get_tree_from_study(study_id='pg_1144', tree_id='tree2324', label_format="name")
+
+