First import of script to extract scansize information from a BNX file.

ef1027e7 · Haarst, Jan van · ef1027e7
Commit ef1027e7 authored Mar 31, 2015 by Haarst, Jan van
--- a/scansize.py
+++ b/scansize.py
+#!/usr/bin/python
+# -*- coding: utf-8 -*-
+"""
+Created on Fri Mar 27 13:27:16 2015
+
+@author: traca001
+"""
+import os
+import sys
+import json
+import fileinput
+import pandas as pd
+import numpy as np
+import matplotlib.pyplot as plt
+
+
+debug = ''
+
+runsizes = dict()
+
+def parse_bnx(file_path, size_dict = {}):
+    for line in fileinput.input():
+        if line.startswith('# Run '):
+            if debug : print line.strip()
+            line = line.split('\t')
+            # The SourceFolder of this Run is the 2nd entry of the line
+            if debug : SourceFolder = line[1]
+            print SourceFolder
+            # The RunId of this Run is the last entry of the line
+            RunId = int(line[-1])
+            if debug : print RunId
+            runsizes[RunId]=dict()
+            # The NumberofScans of this Run is the 8th entry of the line
+            NumberofScans = int(line[7])
+            for Scan in range(1, NumberofScans + 1):
+                runsizes[RunId][Scan] = 0
+                size_dict[Scan] = 0
+            if debug : print runsizes
+            if debug : print size_dict
+        elif line.startswith('0\t'):
+            line = line.split('\t')
+            if debug : print line
+            # Length is the 3rd entry
+            Length = float(line[2])
+            # ScanNumber is the 8th entry
+            ScanNumber = int(line[7])
+            # RunId is the second to last entry
+            RunId = int(line[-2])
+            if debug: print ScanNumber, Length
+            runsizes[RunId][ScanNumber] += Length
+            size_dict[ScanNumber] += Length
+            if debug : print runsizes
+            if debug : print size_dict
+    return size_dict
+
+def hist_plot(size_dict, values = []):
+    for n in size_dict:
+        values.append(size_dict[n])
+    plt.plot(values)
+    plt.show()
+
+if __name__ == '__main__':
+    parse_bnx(sys.argv[1])
+    # print size_dict
+    print json.dumps(runsizes,sort_keys=True,indent=4, separators=(',', ': '))
+    labels = runsizes.keys() # 1,2,..
+
+    # hist_plot(size_dict)
+
+"""
+
+Example data :
+
+# BNX File Version: 1.2
+# Label Channels:   1
+# Nickase Recognition Site 1:
+# Min Molecule Length (Kb): 150
+# Label SNR Filter Type:    Static
+# Min Label SNR:    2.750
+# Software Version: 2.1.0.5973
+#rh SourceFolder    InstrumentSerial    Time    NanoChannelPixelsPerScan    StretchFactor   BasesPerPixel   NumberofScans   ChipId  Flowcell    LabelSNRFilterType  MinMoleculeLength   MinLabelSNR RunId
+# Run data  Z:\Labdata\AlphaUnit_07\2014-10\92544_tomato_nuclei_2014-10-02_10_24\Detect Molecules   ALPHAUNIT07 10/2/2014 10:24:31 AM   68819821    0.85    490.646636962891    12  20249,11887,8/27/2014,850015130 1   Dynamic 100 4.481689    1
+# Run Data  Z:\Labdata\AlphaUnit_07\2014-10\92544_tomato_nuclei_2014-10-01_10_17\Detect Molecules   ALPHAUNIT07 10/1/2014 10:17:14 AM   68819821    0.85    490.646636962891    3   20249,11887,8/27/2014,850015130 1   Dynamic 100 3.669297    2
+# Run Data  Z:\Labdata\AlphaUnit_07\2014-10\92544_tomato_nuclei_2014-10-02_17_40\Detect Molecules   ALPHAUNIT07 10/2/2014 5:40:56 PM    68819821    0.85    490.646636962891    30  20249,11887,8/27/2014,850015130 1   Dynamic 100 4.0552  3
+# Run Data  Z:\Labdata\AlphaUnit_07\2014-10\92544_tomato_nuclei_2014-10-07_09_52\Detect Molecules   ALPHAUNIT07 10/7/2014 9:52:50 AM    68819821    0.85    490.646636962891    30  20249,11887,8/27/2014,850015138 1   Dynamic 100 4.481689    4
+# Run Data  Z:\Labdata\AlphaUnit_07\2014-10\92544_tomato_nuclei_2014-10-07_15_57\Detect Molecules   ALPHAUNIT07 10/7/2014 3:57:36 PM    68819821    0.85    490.646636962891    30  20249,11887,8/27/2014,850015138 1   Dynamic 100 4.481689    5
+# Run Data  Z:\Labdata\AlphaUnit_07\2014-10\92544_tomato_nuclei_2014-10-08_18_51\Detect Molecules   ALPHAUNIT07 10/8/2014 6:51:47 PM    68819821    0.85    490.646636962891    30  20249,11897,9/19/2014,850015323 1   Dynamic 100 4.953032    6
+# Run Data  Z:\Labdata\AlphaUnit_04\2014-10\92544_tomato_nuclei_2014-10-09_16_33\Detect Molecules   ALPHAUNIT04 10/9/2014 4:33:42 PM    68819821    0.85    490.646636962891    30  20249,11897,9/19/2014,850015349 1   Static  0   0   7
+# Run Data  foo\swap92571_tomato_nuclei_2014-10-23_17_27\bar    B023    10/23/2014 5:27:39 PM   69207771    0.85    560.93212890625 30  20249,11887,8/27/2014,850015110 1   Static  0   0   8
+# Run Data  foo\swap92571_tomato_nuclei_2014-10-23_18_22\bar    B023    10/23/2014 6:22:45 PM   69207771    0.85    560.93212890625 30  20249,11887,8/27/2014,850015110 2   Static  0   0   9
+# Quality Score QX01:   SNR
+# Quality Score QX02:   Ave Intensity
+#0h LabelChannel    MoleculeId  Length  AvgIntensity    SNR NumberofLabels  OriginalMoleculeId  ScanNumber  ScanDirection   ChipId  Flowcell    RunId   GlobalScanNumber
+#0f int int float   float   float   int int int int string  int int int
+#1h LabelChannel    LabelPositions[N]
+#1f int float
+#2h LabelChannel    LabelPositions[N]
+#2h int float
+#Qh QualityScoreID  QualityScores[N]
+#Qf str float
+0   3   286801.4    0.177417    42.720  23  3   1   -1  20249,11887,8/27/2014,850015130 1   1   1
+1   12681.6 34673.8 41179.9 44849.6 53791.5 67825.2 69610.3 77348.8 82893.8 88369.7 106001.2    112227.1    133172.5    141298.1    161661.7    169396.6    180201.0    212370.5    221516.8    245870.2    248775.1    276038.5    284017.0    286801.4
+QX11    37.3425 33.6353 10.9667 29.1389 18.0582 30.1201 19.3122 13.6342 25.4127 27.4380 40.7450 22.8699 29.0980 11.4947 57.1931 8.3590  64.5878 23.0364 52.9536 18.7634 58.5257 22.7799 31.2551
+QX12    0.0841  0.0816  0.0458  0.0708  0.0555  0.1042  0.0725  0.0519  0.0638  0.0674  0.1016  0.0647  0.0727  0.0408  0.1263  0.0393  0.1502  0.0704  0.1481  0.0764  0.1518  0.0577  0.0916
+0   4   209794.0    0.117181    20.013  13  4   1   -1  20249,11887,8/27/2014,850015130 1   1   1
+1   11480.0 13296.0 17692.7 91294.8 100287.0    102576.2    124496.7    142540.4    143437.5    173266.5    178092.0    201619.3    206410.0    209794.0
+QX11    8.3894  32.0272 16.9435 70.4016 42.7193 4.9234  29.2025 16.6782 25.7460 16.2417 24.1204 42.1043 8.1092
+QX12    0.0467  0.1062  0.0561  0.1684  0.1221  0.0425  0.0632  0.0575  0.0709  0.0515  0.0655  0.1128  0.0391
+"""
\ No newline at end of file