step6 batch

tf2 · Oct 26, 2023 · eac747b · eac747b
1 parent 1db9f01
commit eac747b
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 21 deletions.
diff --git a/src/Rbin_1.1.tar.gz b/src/Rbin_1.1.tar.gz
diff --git a/src/cnest_dev.py b/src/cnest_dev.py
@@ -222,8 +222,9 @@ def step5 (index_tab, gender_file, rbin_dir, cor_dir, bin_dir, batch_size, targe
         print(cmd5)
         logger.debug('CMD=' + " ".join(cmd5))
         run_cmd(cmd5)
-        logger.info('get_references_to_rbin done')
-        logger.info('Step5 done')
+
+    logger.info('get_references_to_rbin done')
+    logger.info('Step5 done')
 
 
 def step6(rbin_dir, cor_dir, cnv_dir, sample_name, index_tab, cov_file, gender_file, batch_size, cov_cut, cor_cut, skip_em, debug):
@@ -239,15 +240,20 @@ def step6(rbin_dir, cor_dir, cnv_dir, sample_name, index_tab, cov_file, gender_f
         ## updated version
         Rscript /resources/run.R run_hmm_rbin ${rbin_dir} ${cor_dir} ${cnv_dir} ${sample_name} \
             ${index_tab} ${cov_file} ${gender_file} ${batch_size}
+            
+    index_file, rbin_dir, cor_dir, cnv_dir, gender_file, cov_file, batch_size=1000, cov_cut = 20, cor_cut = 0.9, skip_em=FALSE
+
     """
-    logger.info('Start step5')
-    cmd7 = ['Rscript', '/resources/run.R',
-            'run_hmm_rbin', rbin_dir, cor_dir, cnv_dir, sample_name, index_tab,
-            cov_file, gender_file, str(batch_size), str(cov_cut), str(cor_cut), str(skip_em)]
-    print(cmd7)
-    logger.debug('CMD | ' + " ".join(cmd7))
-    run_cmd(cmd7)
-    logger.info('Step5 done')
+    logger.info('Start step6')
+        for i in range(start_pos, start_pos+target_size):
+        cmd7 = ['Rscript', '/resources/run.R',
+                'run_hmm_rbin', index_tab, rbin_dir, cor_dir, cnv_dir, gender_file, cov_file, 
+                str(batch_size), str(target_size), str(start_pos), str(cov_cut), str(cor_cut), str(skip_em)]
+        print(cmd7)
+        logger.debug('CMD | ' + " ".join(cmd7))
+        run_cmd(cmd7)
+
+    logger.info('Step6 done')
 
 
 def ix2id(sample_index, gender_file):

diff --git a/src/run.R b/src/run.R
@@ -136,23 +136,25 @@ if(args[1] == "processLogRtoBin") {
 if(args[1] == "run_hmm_rbin") {
 	#   Rscript /resources/run.R run_hmm_rbin ${bin_dir} ${cor_dir} ${cnv_dir} ${sample_name} \
     #        ${index_tab} ${cov_file} ${gender_file} ${batch_size}
-	rbin_dir = args[2]
-	cor_dir = args[3]
-	cnv_dir = args[4]
-	sample_name = args[5]
-	index_file = args[6]
+    # index_file, rbin_dir, cor_dir, cnv_dir, gender_file, cov_file, batch_size=1000, target_size, start_pos, cov_cut = 20, cor_cut = 0.9, skip_em=FALSE
+	index_file = args[2]
+	rbin_dir = args[3]
+	cor_dir = args[4]
+	cnv_dir = args[5]
+	gender_file = args[6]
 	cov_file = args[7] # cov_file=/resources/txt_covered_all_coverage.txt 
-	gender_file = args[8]
-	batch_size = args[9]
-    cov_cut = as.numeric(as.character(args[10]))
-    cor_cut = as.numeric(as.character(args[11]))
-    skip_em = as.character(args[12])
+	batch_size = as.numeric(as.character(args[8]))
+	target_size = as.numeric(as.character(args[9]))
+	start_pos = as.numeric(as.character(args[10]))
+    cov_cut = as.numeric(as.character(args[11]))
+    cor_cut = as.numeric(as.character(args[12]))
+    skip_em = as.character(args[13])
     if(skip_em=="True") {
         skip_em = TRUE
     } else {
         skip_em = FALSE
     }
 	print(sample_name)
-	run_hmm_rbin(rbin_dir, sample_name, index_file, cov_file, cor_dir, gender_file, cnv_dir, batch_size, cov_cut, cor_cut, skip_em)
+	run_hmm_rbin(index_file, rbin_dir, cor_dir, cnv_dir, gender_file, cov_file, batch_size, target_size, start_pos, cov_cut, cor_cut, skip_em)
 	print ("run_hmm_rbin")
 }