From 9690d1ae0e9ea7c9a63336b459f3c931381f7e68 Mon Sep 17 00:00:00 2001
From: zhangrengang <zhangrenang@ori-gene.cn>
Date: Wed, 25 Sep 2024 21:17:15 +0800
Subject: [PATCH] add SOI-tools

---
 SOI-tools.md     | 25 +++++++++++++++++++++++++
 setup.py         |  1 +
 soi/RunCmdsMP.py |  6 ++++--
 soi/mcscan.py    |  2 +-
 soi/options.py   |  3 ++-
 5 files changed, 33 insertions(+), 4 deletions(-)
 create mode 100644 SOI-tools.md

diff --git a/SOI-tools.md b/SOI-tools.md
new file mode 100644
index 0000000..c2f9851
--- /dev/null
+++ b/SOI-tools.md
@@ -0,0 +1,25 @@
+
+#### Macro-synteny phylogeny ####
+Before run the pipeline:
+1. install the lasest verion of [SOI](https://github.com/zhangrengang/orthoindex#installation) (not the release version);
+2. completed the [example pipeline] (https://github.com/zhangrengang/evolution_example) to get the orthologous synteny file `collinearity.ortho`;
+3. prepare file `species.config` to set the expected subgenome numbers for targeted species (TAB seperated):
+```
+Vitis_vinifera  1
+Aralia_elata    2
+Centella_asiatica       2
+```
+Then run the pipeline:
+```
+cd phylogenomics
+soi-syn anchor_trees collinearity.ortho species.config ../pep.faa ../all_species_gene.gff output_dir
+```
+After the pipeline completed, you can find tree files in the `output_dir`:
+```
+OG*treefile		# gene tree file for each anchor gene
+chr*treefile	# macro-synteny tree file by concatenating the anchor genes from the same chromosome set
+CHR*treefile	# macro-synteny tree file, but allowing gene missing
+```
+For example, `CHR_Ae11-Ae15-Ca2-Ca7-Vv1_143_283.concat.treefile`, `Ae11-Ae15-Ca2-Ca7-Vv1` is the chromosome set,
+`143` is the number of anchor genes, and `283` is the number of all syntenic genes allowing missing.
+
diff --git a/setup.py b/setup.py
index 120b7a9..50217b4 100644
--- a/setup.py
+++ b/setup.py
@@ -23,6 +23,7 @@
     scripts=[],
     entry_points={
         'console_scripts': ['soi = soi.options:main',
+							'soi-syn = soi.mcscan:main',
         ],
     },
 )
diff --git a/soi/RunCmdsMP.py b/soi/RunCmdsMP.py
index f833dd5..e693484 100644
--- a/soi/RunCmdsMP.py
+++ b/soi/RunCmdsMP.py
@@ -26,9 +26,11 @@
 	from tempfile import NamedTemporaryFile
 except (RuntimeError,ImportError,AttributeError,OSError) as e:
 	if "DRMAA_LIBRARY_PATH" in format(e):
-		logger.warning('Grid computing is not available because DRMAA not configured properly: {}'.format(e))
+#		logger.warning('Grid computing is not available because DRMAA not configured properly: {}'.format(e))
+		pass
 	else:
-		logger.warning('Grid computing is not available because DRMAA not installed: {}'.format(e))
+#		logger.warning('Grid computing is not available because DRMAA not installed: {}'.format(e))
+		pass
 	logger.info('No DRMAA (see https://github.com/pygridtools/drmaa-python), Switching to local mode.')
 	GRID = False
 
diff --git a/soi/mcscan.py b/soi/mcscan.py
index bc00509..b3711ac 100644
--- a/soi/mcscan.py
+++ b/soi/mcscan.py
@@ -2226,7 +2226,7 @@ def chrom_trees(self, min_genes=2):
 				continue
 			xxchroms += [chroms]
 			i += 1
-			prefix = '-'.join(chroms) + '_' + str(len(alnfiles))
+			prefix = 'chr_' + '-'.join(chroms) + '_' + str(len(alnfiles))
 			
 			cmds = self.concat_tree(alnfiles, prefix, idmap=self.d_chroms, astral=True)
 		#	d_gene_count3[concat_alnfile] = len(alnfiles)
diff --git a/soi/options.py b/soi/options.py
index 2f350db..a9be961 100644
--- a/soi/options.py
+++ b/soi/options.py
@@ -216,7 +216,8 @@ def main():
 	key = sys.argv[1]
 	func = FUNC[key]
 	func(**args.__dict__)
-		
+	print()
+	
 if __name__ == '__main__':
 	main()