Add N90 and 5 longest reads #113

workflows/ont_qc/components/run_stats.py

@@ -72,22 +72,40 @@ class Run_stats (Analysis):
             if parts[0] == "total_bases": stats["total_bases"] = parts[1]
             if parts[0] == "mean_read_length": stats["mean_read_length"] = parts[1]
             if parts[0] == "N50_read_length": stats["N50_read_length"] = parts[1]
+            if parts[0] == "N90_read_length": stats["N90_read_length"] = parts[1]
             if parts[0] == "mean_read_quality": stats["mean_read_quality"] = parts[1]
             if parts[0] == "mean_yield_per_sec": stats["mean_yield_per_sec"] = parts[1]
+            if parts[0] == "longread1": stats["longread1"] = parts[1]
+            if parts[0] == "longread2": stats["longread2"] = parts[1]
+            if parts[0] == "longread3": stats["longread3"] = parts[1]
+            if parts[0] == "longread4": stats["longread4"] = parts[1]
+            if parts[0] == "longread5": stats["longread5"] = parts[1]
 
             if parts[0] == "nb_reads_Q7": stats["nb_reads_Q7"] = parts[1]
             if parts[0] == "total_bases_Q7": stats["total_bases_Q7"] = parts[1]
             if parts[0] == "mean_read_length_Q7": stats["mean_read_length_Q7"] = parts[1]
             if parts[0] == "N50_read_length_Q7": stats["N50_read_length_Q7"] = parts[1]
+            if parts[0] == "N90_read_length_Q7": stats["N90_read_length_Q7"] = parts[1]
             if parts[0] == "mean_read_quality_Q7": stats["mean_read_quality_Q7"] = parts[1]
             if parts[0] == "mean_yield_per_sec_Q7": stats["mean_yield_per_sec_Q7"] = parts[1]
+            if parts[0] == "longread1_Q7": stats["longread1_Q7"] = parts[1]
+            if parts[0] == "longread2_Q7": stats["longread2_Q7"] = parts[1]
+            if parts[0] == "longread3_Q7": stats["longread3_Q7"] = parts[1]
+            if parts[0] == "longread4_Q7": stats["longread4_Q7"] = parts[1]
+            if parts[0] == "longread5_Q7": stats["longread5_Q7"] = parts[1]
             
             if parts[0] == "nb_reads_Q9": stats["nb_reads_Q9"] = parts[1]
             if parts[0] == "total_bases_Q9": stats["total_bases_Q9"] = parts[1]
             if parts[0] == "mean_read_length_Q9": stats["mean_read_length_Q9"] = parts[1]
             if parts[0] == "N50_read_length_Q9": stats["N50_read_length_Q9"] = parts[1]
+            if parts[0] == "N90_read_length_Q9": stats["N90_read_length_Q9"] = parts[1]
             if parts[0] == "mean_read_quality_Q9": stats["mean_read_quality_Q9"] = parts[1]
             if parts[0] == "mean_yield_per_sec_Q9": stats["mean_yield_per_sec_Q9"] = parts[1]
+            if parts[0] == "longread1_Q9": stats["longread1_Q9"] = parts[1]
+            if parts[0] == "longread2_Q9": stats["longread2_Q9"] = parts[1]
+            if parts[0] == "longread3_Q9": stats["longread3_Q9"] = parts[1]
+            if parts[0] == "longread4_Q9": stats["longread4_Q9"] = parts[1]
+            if parts[0] == "longread5_Q9": stats["longread5_Q9"] = parts[1]
             
         #print(stats)
         return stats
@@ -142,36 +160,57 @@ class Run_stats (Analysis):
                     
             group = 'rawdata'
             metrics.append(group)
-            self._add_result_element("metrics", "headers", ','.join(["nb_reads", "total_bases", "mean_read_length","N50_read_length", "mean_read_quality", "mean_yield_per_sec"]), group)
-            
-            self._add_result_element(sample, "nb_reads", str(stat_info["nb_reads"]),group),
-            self._add_result_element(sample, "total_bases", str(stat_info["total_bases"]),group),
-            self._add_result_element(sample, "mean_read_length", str(stat_info["mean_read_length"]),group),
-            self._add_result_element(sample, "N50_read_length", str(stat_info["N50_read_length"]),group),
-            self._add_result_element(sample, "mean_read_quality", str(stat_info["mean_read_quality"]),group),
-            self._add_result_element(sample, "mean_yield_per_sec", str(stat_info["mean_yield_per_sec"]),group),
+            metrics_names = ["nb_reads", 
+                            "total_bases", 
+                            "mean_read_length",
+                            "N50_read_length",
+                            "N90_read_length",
+                            "mean_read_quality",
+                            "mean_yield_per_sec",
+                            "longread1",
+                            "longread2",
+                            "longread3",
+                            "longread4",
+                            "longread5"]
+            self._add_result_element("metrics", "headers", ','.join(metrics_names), group)
+            for item in metrics_names:
+                self._add_result_element(sample, item, str(stat_info[item]),group)
             
             group = 'Q7'
             metrics.append(group)
-            self._add_result_element("metrics", "headers", ','.join(["nb_reads_Q7", "total_bases_Q7", "mean_read_length_Q7","N50_read_length_Q7", "mean_read_quality_Q7", "mean_yield_per_sec_Q7"]), group)
-            
-            self._add_result_element(sample, "nb_reads_Q7", str(stat_info["nb_reads_Q7"]),group),
-            self._add_result_element(sample, "total_bases_Q7", str(stat_info["total_bases_Q7"]),group),
-            self._add_result_element(sample, "mean_read_length_Q7", str(stat_info["mean_read_length_Q7"]),group),
-            self._add_result_element(sample, "N50_read_length_Q7", str(stat_info["N50_read_length_Q7"]),group),
-            self._add_result_element(sample, "mean_read_quality_Q7", str(stat_info["mean_read_quality_Q7"]),group),
-            self._add_result_element(sample, "mean_yield_per_sec_Q7", str(stat_info["mean_yield_per_sec_Q7"]),group),
+            metrics_names = ["nb_reads_Q7",
+                            "total_bases_Q7",
+                            "mean_read_length_Q7",
+                            "N50_read_length_Q7",
+                            "N90_read_length_Q7",
+                            "mean_read_quality_Q7",
+                            "mean_yield_per_sec_Q7",
+                            "longread1_Q7",
+                            "longread2_Q7",
+                            "longread3_Q7",
+                            "longread4_Q7",
+                            "longread5_Q7"]
+            self._add_result_element("metrics", "headers", ','.join(metrics_names), group)
+            for item in metrics_names:
+                self._add_result_element(sample, item, str(stat_info[item]),group)
             
             group = 'Q9'
             metrics.append(group)
-            self._add_result_element("metrics", "headers", ','.join(["nb_reads_Q9", "total_bases_Q9", "mean_read_length_Q9","N50_read_length_Q9", "mean_read_quality_Q9", "mean_yield_per_sec_Q9"]), group)
-            
-            self._add_result_element(sample, "nb_reads_Q9", str(stat_info["nb_reads_Q9"]),group),
-            self._add_result_element(sample, "total_bases_Q9", str(stat_info["total_bases_Q9"]),group),
-            self._add_result_element(sample, "mean_read_length_Q9", str(stat_info["mean_read_length_Q9"]),group),
-            self._add_result_element(sample, "N50_read_length_Q9", str(stat_info["N50_read_length_Q9"]),group),
-            self._add_result_element(sample, "mean_read_quality_Q9", str(stat_info["mean_read_quality_Q9"]),group),
-            self._add_result_element(sample, "mean_yield_per_sec_Q9", str(stat_info["mean_yield_per_sec_Q9"]),group),
+            metrics_names = ["nb_reads_Q9",
+                            "total_bases_Q9",
+                            "mean_read_length_Q9",
+                            "N50_read_length_Q9",
+                            "N90_read_length_Q9",
+                            "mean_read_quality_Q9",
+                            "mean_yield_per_sec_Q9",
+                            "longread1_Q9",
+                            "longread2_Q9",
+                            "longread3_Q9",
+                            "longread4_Q9",
+                            "longread5_Q9"]
+            self._add_result_element("metrics", "headers", ','.join(metrics_names), group)
+            for item in metrics_names:
+                self._add_result_element(sample, item, str(stat_info[item]),group)
             
         group = 'plots'
         metrics.append(group)

workflows/ont_qc/data/ont_qc.conf

@@ -16,7 +16,7 @@
 #
 
 --date
-31/07/2017
+23/07/2019
 --data-nature
 DNA
 --sequencer
@@ -28,10 +28,14 @@ Unknown
 --species
 arabido
 --description
-ont demo workflow
+"ont demo workflow, Minknow 1.10 and Albacore 2.0.2"
 --project-name
-test-dev
+REPDEV
 
+--summary-file
+/save/ng6-test/src/ng6-git-mmanno/workflows/ont_qc/data/sequencing_summary_barcoded_test.txt
+--barcoded
+no
 --sample
 sample-name=ONT_sample
-read1=/MinION/20170308_arab1D_LSK108_MN19767/20170308_arab1D_LSK108_MN19767_albacorev1.2.4/20170308_arab1D_LSK108_MN19767_albacorev1.2.4.fastq
+read1=/save/ng6-test/src/ng6-git-mmanno/workflows/ont_qc/data/test.fastq