re-implement nonltr

MGEScan · Feb 17, 2016 · 8193304 · 8193304
1 parent e68403c
commit 8193304
Show file tree

Hide file tree

Showing 2 changed files with 57 additions and 7 deletions.
diff --git a/mgescan/nonltr/hmm/hmm_lib.c b/mgescan/nonltr/hmm/hmm_lib.c
@@ -702,12 +702,14 @@ void get_hydro(int start, int end, double *score, char *O){
 	}
 	memset(seq, '\0', 20000);
 
-	strcpy(temp_file1, out_dir);
+	//strcpy(temp_file1, out_dir);
+	strcpy(temp_file1, "/tmp/");
 	strcat(temp_file1, "ppppp.XXXXXX");
 	int fd1;
 	fd1 = mkstemp(temp_file1);
 
-	strcpy(temp_file2, out_dir);
+	//strcpy(temp_file2, out_dir);
+	strcpy(temp_file2, "/tmp/");
 	strcat(temp_file2, "qqqqq.XXXXXX");
 	int fd2;
 	fd2 = mkstemp(temp_file2);

diff --git a/mgescan/nonltr/nonltr.py b/mgescan/nonltr/nonltr.py
@@ -21,12 +21,17 @@ class NonLTR(object):
     ver = 'MGEScan NonLTR 3.0.0'
     mpi_cmd = "mpi_mgescan"
     hmmerv = "3"
-    tag_for_backward = "_b"
 
     def __init__(self):
         self.set_inputs()
         self.get_env()
         self.set_param()
+        self.set_cmd()
+
+    def set_cmd(self):
+        self.run_hmm_cmd = self.base_path + "nonltr/" + "run_hmm.pl"
+        self.run_post_cmd = self.base_path + "nonltr/" + "post_process.pl"
+        self.run_post2_cmd = self.base_path + "nonltr/" + "post_process2.pl"
 
     def get_env(self):
         self.base_path = os.environ.get("MGESCAN_SRC") + "/mgescan/"
@@ -38,8 +43,9 @@ def set_param(self):
         self.p_prg = "--prg " + self.name
         self.p_mgescan_mpi_cmd = self.base_path + self.mpi_cmd
         self.p_genome_f = "--genome " + self.genome_path
-        self.p_genome_b = "--genome " + self.genome_path + self.tag_for_backward
-        self.p_data = "--data " + self.output_path
+        self.p_genome_b = "--genome " + self.genome_path + "_b"
+        self.p_data_f = "--data " + self.output_path + "/f/"
+        self.p_data_b = "--data " + self.output_path + "/b/"
         self.p_hmmerv = "--hmmerv " + self.hmmerv
         self.set_mpi_option()
 
@@ -54,18 +60,57 @@ def read_mpi_host_file(self):
         self.p_hf_option = ""
 
     def forward(self):
+        utils.create_directory(self.output_path, False)
         if self.nmpi:
             cmd = self._padding("mpirun", self.p_np, self.p_mpi_option,
                     self.p_mgescan_mpi_cmd, self.p_prg, self.p_genome_f,
-                    self.p_data, self.p_hmmerv)
+                    self.p_data_f, self.p_hmmerv)
             self.run_cmd(cmd)
+        else:
+           from os import listdir
+           from os.path import isfile, join
+           mypath = self.genome_path
+           for f in listdir(mypath):
+               fpath = join(mypath, f)
+               if isfile(fpath):
+                   p = Process(target=self.run_hmm, args=("forward", fpath,))
+                   p.start()
+        self.post_process("forward")
+
+    def post_process(self, t):
+        if t == "forward":
+            cmd = self.run_post_cmd + " --dna=" + self.genome_path + " --out=" + self.output_path + "/f/" + " --rev=0"
+        elif t == "backward":
+            cmd = self.run_post_cmd + " --dna=" + self.genome_path + " --out=" + self.output_path + "/b/" + " --rev=1"
+        self.run_cmd(cmd)
+
+    def run_hmm(self, t, fname):
+        if t == "forward":
+            cmd = self.run_hmm_cmd + " --dna=" + fname + " --out=" + self.output_path + "/f/" + " --hmmerv=" + self.hmmerv
+        elif t == "backward":
+            cmd = self.run_hmm_cmd + " --dna=" + fname + " --out=" + self.output_path + "/b/" + " --hmmerv=" + self.hmmerv
+        self.run_cmd(cmd)
+
+    def post_process2(self):
+        cmd = self.run_post2_cmd + " --data_dir=" + self.output_path + " --hmmerv=" + self.hmmerv
+        self.run_cmd(cmd)
 
     def backward(self):
         if self.nmpi:
             cmd = self._padding("mpirun", self.p_np, self.p_mpi_option,
                     self.p_mgescan_mpi_cmd, self.p_prg, self.p_genome_b,
-                    self.p_data, self.p_hmmerv)
+                    self.p_data_b, self.p_hmmerv)
             self.run_cmd(cmd)
+        else:
+           from os import listdir
+           from os.path import isfile, join
+           mypath = self.genome_path + "_b"
+           for f in listdir(mypath):
+               fpath = join(mypath, f)
+               if isfile(fpath):
+                   p = Process(target=self.run_hmm, args=("backward",fpath,))
+                   p.start()
+        self.post_process("backward")
 
     def _padding(self, *args):
         res = ""
@@ -75,6 +120,7 @@ def _padding(self, *args):
             return res[1:]
         else:
             return res
+
     def run_cmd(self, cmd):
         print cmd
         return
@@ -105,6 +151,8 @@ def run(self):
         if 'p2' in locals():
             p2.join()
 
+        self.post_process2()
+
 if __name__ == '__main__':
     obj = NonLTR()
     obj.run()