ENH include density in the percontigs output

BigDataBiology · Jul 13, 2023 · e17335e · e17335e
1 parent 60b2b47
commit e17335e
Show file tree

Hide file tree

Showing 9 changed files with 10 additions and 4 deletions.
diff --git a/install.sh b/install.sh
@@ -43,7 +43,7 @@ fi
 ${CONDA_INSTALL_CMD} install -y \
           --prefix $BASEDIR/envs/Macrel_env \
           ngless \
-          pyrodigal>=0.7.3 \
+          "pyrodigal>=0.7.3" \
           megahit \
           paladin \
           pandas \

diff --git a/macrel/data/models/AMP.pkl.gz b/macrel/data/models/AMP.pkl.gz
diff --git a/macrel/data/models/Hemo.pkl.gz b/macrel/data/models/Hemo.pkl.gz
diff --git a/macrel/main.py b/macrel/main.py
@@ -294,13 +294,14 @@ def do_density(args, clen, prediction):
     clen = clen.merge(on='contig', right=tpred, how='outer').fillna(0)
     clen[clen.columns[1:]] = clen[clen.columns[1:]].astype(int)   
     ofile = path.join(args.output, args.outtag + '.percontigs.gz')
+    sample = clen.set_index('contig').sum(axis=0).tolist()
+    sample_density = sample[-1] * 1e6 / sample[0]
     with open_output(ofile, mode='wb') as raw_out:
         with gzip.open(raw_out, 'wt') as out:
-            from macrel_version import __version__
+            from .macrel_version import __version__
             out.write('# Prediction from macrel v{}\n'.format(__version__))
+            out.write(f'# Macrel calculated for the sample a density of {sample_density:.3f} AMPs / Mbp.\n')
             clen.to_csv(out, sep='\t', index=False, float_format="%.3f")
-    sample = clen.set_index('contig').sum(axis=0).tolist()
-    sample_density = sample[-1] * 1e6 / sample[0]
     print(f'Macrel processed the sample and verified a density of {sample_density:.3f} AMPs / Mbp.')
 
 def do_get_examples(args):

diff --git a/mkdocs.yml b/mkdocs.yml
@@ -23,6 +23,7 @@ nav:
     - 'Macrel': index.md
     - 'Install': install.md
     - 'Usage': usage.md
+    - 'Tutorial': tutorial.md
     - "What's New (history)": whatsnew.md
     - 'Contacts': contact.md
     - 'FAQ': faq.md

diff --git a/tests/contigs.nosmorfs/expected.percontigs b/tests/contigs.nosmorfs/expected.percontigs
@@ -1,3 +1,4 @@
 # Prediction from macrel v1.2.0
+# Macrel calculated for the sample a density of 0.000 AMPs / Mbp.
 contig	length	ORFs	smORFs	AMPs
 scaffold2530_2_MH0058	1324	1	0	0
diff --git a/tests/contigs/expected.percontigs b/tests/contigs/expected.percontigs
@@ -1,4 +1,5 @@
 # Prediction from macrel v1.2.0
+# Macrel calculated for the sample a density of 45.062 AMPs / Mbp.
 contig	length	ORFs	smORFs	AMPs
 scaffold2530_2_MH0058	717	2	2	0
 scaffold75334_1_MH0058	3424	1	1	1

diff --git a/tests/reads.se/expected.percontigs b/tests/reads.se/expected.percontigs
@@ -1,4 +1,5 @@
 # Prediction from macrel v1.2.0
+# Macrel calculated for the sample a density of 59.743 AMPs / Mbp.
 contig	length	ORFs	smORFs	AMPs
 k47_0	3379	4	2	0
 k47_1	6046	9	4	0

diff --git a/tests/reads/expected.percontigs b/tests/reads/expected.percontigs
@@ -1,4 +1,5 @@
 # Prediction from macrel v1.2.0
+# Macrel calculated for the sample a density of 57.627 AMPs / Mbp.
 contig	length	ORFs	smORFs	AMPs
 k77_3	1270	1	0	0
 k77_5	5202	6	3	0