Support for timTOF - pt1

* rewritten routines to be able to work with different input formats
* opening of spectrum works
* lots of things broken on the way. Now its time to fix and cleanup

prasopes/__main__.py

@@ -24,6 +24,7 @@ import prasopes.graphtools as gt
 import prasopes.imagetools as imgt
 import prasopes.zcetools as zce
 import prasopes.docks as docks
+import prasopes.datasets as datasets
 import prasopes.tangoicons
 import sys
 import logging
@@ -64,12 +65,15 @@ def show_exception_and_exit(exc_type, exc_value, tb):
 
 
 def load_file(parent, augCanvas, update, settings, loadthread, filename=None,
-              nothreading=False):
+              nothreading=True):
     """populates dataset and plots it"""
-    directory = augCanvas.filename or settings.value("open_folder")
+    directory = augCanvas.ds.filename\
+            if not (isinstance(augCanvas.ds, list))\
+            else settings.value("open_folder")
     filename = filename or QtWidgets.QFileDialog.getOpenFileName(
             caption="Open spectrum", directory=directory,
-            filter="Finnigan RAW files (*.raw, *.RAW)")[0]
+            filter="All supported formats (*.raw *.RAW *.d);;"
+                   "Finnigan RAW files (*.raw *.RAW);;timsTOF file (*.d)")[0]
     if filename != '' and os.path.isfile(filename)\
             and not os.path.isdir(filename):
         error = update_signal()
@@ -77,12 +81,9 @@ def load_file(parent, augCanvas, update, settings, loadthread, filename=None,
 
         def runfnc():
             try:
-                [i.clear() for i in (
-                    augCanvas.ds, augCanvas.chrom['timesarg'],
-                    augCanvas.ms['params'], augCanvas.ms['headers'])]
-                [augCanvas.ds.append(dict(
-                    chrom_dat=i[0], masses=i[1], matrix=i[2]))
-                 for i in load_raw(filename, settings.value("tmp_location"))]
+                augCanvas.ds = datasets.ThermoRawDataset(filename) if\
+                        os.path.splitext(filename)[1] in (".raw", ".RAW")\
+                        else datasets.BrukerTimsDataset(filename)
             except rawprasslib.ParsingException as pex:
                 errormsg.append("Opening of the file has failed!")
                 errormsg.append(
@@ -91,29 +92,10 @@ def load_file(parent, augCanvas, update, settings, loadthread, filename=None,
                     "Error message:\n{}".format(pex.args[0]))
                 error.signal.emit()
                 return
-            if autoparams:
-                try:
-                    (augCanvas.ms['params'], rawheaders,
-                        augCanvas.chrom['machtype']) = load_params(
-                                filename, settings.value("tmp_location"))
-                    segments = [len(subset['chrom_dat'][0])
-                                for subset in augCanvas.ds]
-                    indicies = [sum(segments[:i+1])
-                                for i in range(len(segments))]
-                    augCanvas.ms['headers'] = np.split(
-                            rawheaders, indicies)[:-1]
-                except Exception as pex:
-                    errormsg.append(
-                            "File is incompatible with the rawautoparams,")
-                    errormsg.append(
-                            "no parameters loaded!\n\n"
-                            "Error message:\n{}".format(pex.args[0]))
-                    error.signal.emit()
             gt.populate(augCanvas)
-            augCanvas.filename = filename
             oldrecents = settings.value("recents")
             oldrecents.remove(filename) if filename in oldrecents else None
-            settings.setValue("recents", [filename, *oldrecents])
+            settings.setValue("recents", [filename, *oldrecents][:10])
             update.signal.emit()
         error.signal.connect(lambda: QtWidgets.QMessageBox.critical(
             parent, errormsg[0], errormsg[1]))

prasopes/datasets.py

+#!/usr/bin/env python3
+from rawprasslib import load_raw
+from opentimspy.opentims import OpenTIMS
+try:
+    from rawautoparams import load_params
+    import rawautoparams
+    autoparams = True
+except ImportError:
+    autoparams = False
+import pathlib
+import opentims_bruker_bridge
+import numpy as np
+import prasopes.config as cf
+import prasopes.datatools as dt
+import os.path
+
+
+class Dataset():
+    def __init__(self, rawfile):
+        self.filename = rawfile
+        self.chromatograms = []
+        self.dataset = []
+        self.timemin = -np.inf
+        self.timemax = np.inf
+
+    def refresh(self):
+        """implement per-case"""
+        return None
+
+    def get_chromatograms(self):
+        """implement per-case"""
+        raise NotImplementedError
+
+    def get_spectra(self):
+        """implement per-case"""
+        raise NotImplementedError
+
+
+class ThermoRawDataset(Dataset):
+    def __init__(self, rawfile):
+        super().__init__(rawfile)
+        self.params = []
+        self.machtype = []
+        self.headers = []
+        self.refresh()
+
+    def refresh(self):
+        self.dataset = load_raw(self.filename, 
+                                cf.settings().value("tmp_location"))
+        self.chromatograms = self.get_chromatograms()
+        self.timemin, self.timemax = [self.chromatograms[i][0][i] 
+                                      for i in (0,-1)]
+        if autoparams:
+            try:
+                self.params, rawheaders, self.machtype = load_params(
+                    self.filename, cf.settings().value("tmp_location"))
+                segments = [len(i[0]) for i in self.chromatograms]
+                indicies = [sum(segments[:i+1]) for i,j in enumerate(segments)]
+                self.headers = np.split(rawheaders, indicies)[:-1]
+            except Exception as pex:
+                self.params, self.machtype, self.headers = None, None, None
+
+    def get_chromatograms(self):
+        if cf.settings().value("view/oddeven", type=bool):
+            chroms = []
+            for i in self.dataset:
+                for j in (0,1):
+                    chroms.append([i[0][ax, :][j::2] for ax in (0,1)])
+        else:
+            chroms = [i[0] for i in self.dataset]
+        return chroms
+
+    def get_spectra(self):
+        times = dt.argsubselect(np.concatenate(
+             [subset[0][0] for subset in self.dataset]),
+             self.timemin, self.timemax)
+        args = []
+        for subset in self.dataset:
+            goodtimes = np.where((times < len(subset[0][0]))
+                             & ~(times < 0))[0]
+            args.append(times[goodtimes])
+            times =- len(subset[0][0])
+            spectra = []
+            for i,subset in enumerate(self.dataset):
+                if cf.settings().value("view/oddeven", type=bool):
+                    for j in (0,1):
+                        yvalz = np.mean(subset[2][args[i][j::2]], axis=0)
+                        spectra.append([subset[1], yvalz])
+                else:
+                    yvalz = np.mean(subset[2][args[i]], axis=0)
+                    spectra.append([subset[1], yvalz])
+        return spectra
+
+
+class BrukerTimsDataset(Dataset):
+    def __init__(self, rawfile):
+        super().__init__(rawfile)
+        self.refresh()
+
+    def refresh(self):
+        if(os.path.isdir(self.filename)):
+            self.dataset = OpenTIMS(pathlib.Path(self.filename))
+        else:
+            self.dataset = OpenTIMS(pathlib.Path(os.path.dirname(self.filename)))
+        self.chromatograms = self.get_chromatograms()
+        self.timemin, self.timemax = [self.chromatograms[i][0][i] 
+                                      for i in (0,-1)]
+
+    def get_chromatograms(self):
+        keys = ('retention_time', 'intensity')
+        times = self.dataset.retention_times / 60
+        intensities = [np.sum(i['intensity']) for i in self.dataset.query_iter(
+                       self.dataset.ms1_frames, columns=('intensity',))]
+        return [[times,intensities]]
+
+    def get_spectra(self):
+        massints = self.dataset.rt_query(self.timemin*60, self.timemax*60,
+                                               columns=('mz', 'intensity'))
+        sortmasses = np.sort(massints['mz'])
+        masssteps = sortmasses[1:] - sortmasses[:-1]
+        binspos = np.where(masssteps > 0.001)[0]
+        bins = sortmasses[:-1][binspos] + (masssteps[binspos]/2)
+        binpos = np.digitize(massints['mz'], bins)
+        masses = np.bincount(binpos, massints['mz']) / np.bincount(binpos)
+        ints = np.bincount(binpos, massints['intensity']) / self.dataset.max_frame
+        return [[masses, ints]]

prasopes/graphtools.py

@@ -26,6 +26,7 @@ class AugFigureCanvas(FigureCanvas):
         self.figure = Figure(figsize=(5, 4), dpi=100, facecolor="None",
                              constrained_layout=True)
         super().__init__(self.figure)
+        self.filename = ""
         self.ds = []
         self.ms = dict(annotation=[], name="Spectrum", xlabel="m/z",
                        ylabel="ion count", xtics=20, predict=None,
@@ -34,7 +35,6 @@ class AugFigureCanvas(FigureCanvas):
             x=[0], y=[0], t_start=None, t_end=None, machtype=None,
             name="Chromatogram", xlabel="time(min)", ylabel="total ion count",
             timesarg=[])
-        self.filename = None
         self.drlcache = [None, None]
         grid = self.figure.add_gridspec(2, 1)
         self.chromplot = self.figure.add_subplot(grid[0, 0],
@@ -296,7 +296,10 @@ def plot_subtime(augCanvas):
 
 def pick_times(x_min, x_max, augCanvas):
     """subselect part of the chromatogram and plot it"""
-    augCanvas.chrom['t_start'] = x_min
+    augCanvas.ds.timemin = x_min
+    augCanvas.ds.timemax = x_max
+    populate(augCanvas)
+    """augCanvas.chrom['t_start'] = x_min
     augCanvas.chrom['t_end'] = x_max
     times = dt.argsubselect(np.concatenate(
         [subset['chrom_dat'][0] for subset in augCanvas.ds]), x_min, x_max)
@@ -307,7 +310,7 @@ def pick_times(x_min, x_max, augCanvas):
         augCanvas.chrom['timesarg'].append(times[goodtimes])
         times = times - len(subset['chrom_dat'][0])
     update_paramstable(augCanvas)
-    plot_subtime(augCanvas)
+    plot_subtime(augCanvas)"""
 
 
 def shift_times(event, augCanvas):
@@ -408,6 +411,7 @@ def ann_spec(ms_spec, msdata, ann_limit=0.01):
 def pop_plot(xdata, ydata, plot, plot_data, colornum=0, legend=None, annotate=True):
     """Define and populate plot"""
     if len(xdata):
+        print(colornum)
         plot.plot(xdata, ydata, linewidth=1, color=(
             colors[colornum % len(colors)]/255), label=legend)
     plot.set_title(plot_data['name'], loc="right")
@@ -486,14 +490,18 @@ def populate(augCanvas):
         [pints.extend([np.nan, 0, i]) for i in crudeints]
         augCanvas.spectplot.plot(pmasses, pints, linewidth=1)
 
-    augCanvas.chrom['t_start'] = augCanvas.ds[0]['chrom_dat'][0, 0]
-    augCanvas.chrom['t_end'] = augCanvas.ds[-1]['chrom_dat'][0, -1]
-    for i, subset in enumerate(augCanvas.ds):
-        if len(augCanvas.ms['headers']) == len(augCanvas.ds):
-            legend = legendize(augCanvas.ms['headers'][i], augCanvas.chrom)
+    chromxy = augCanvas.ds.get_chromatograms()
+    msxy = augCanvas.ds.get_spectra()
+    for i in range(len(msxy)):
+        #if len(augCanvas.ds.headers) == len(augCanvas.ds.chromatograms):
+        if False:
+            legend = legendize(augCanvas.ds.headers[i], augCanvas.chrom)
         else:
             legend = None
-        if cf.settings().value("view/oddeven", type=bool):
+        pop_plot(msxy[i][0], msxy[i][1], augCanvas.spectplot, augCanvas.ms, i, legend)
+        pop_plot(chromxy[i][0], chromxy[i][1], augCanvas.chromplot,
+                augCanvas.chrom, i, legend)
+        """if cf.settings().value("view/oddeven", type=bool):
             msx = subset['masses']
             for j in range(2):
                 msy = np.mean(subset['matrix'][j::2], axis=0)
@@ -510,7 +518,7 @@ def populate(augCanvas):
             chromy = subset['chrom_dat'][1, :]
             pop_plot(msx, msy, augCanvas.spectplot, augCanvas.ms, i, legend)
             pop_plot(chromx, chromy, augCanvas.chromplot, augCanvas.chrom,
-                     i, legend)
+                     i, legend)"""
     for ax in (augCanvas.spectplot, augCanvas.chromplot):
         if not augCanvas.ms['headers'] == []:
             ax.legend(loc=2)

prasopes/tests/loadrawtest.sh

+#!/bin/bash
+
+prasopes ~/roithPhD/sw_modd/msresearch/bench_batch/TSQ.raw

prasopes/tests/loadtims.sh

+#!/bin/bash
+
+prasopes ~/roithPhD/sw_modd/msresearch/opentims_spektra/210211_PEG600_TIMS_Detect_Calibrated.d/210211_PEG600_TIMS_Detect_Calibrated.d