コード例 #1
0
def singleVRTrialAnalysis(fileToAnalyse):
    # TODO: add savePlots,recomputeFlyData as function arguments

    # ------------------------------------------------------------------------------------------------------------------
    # Extract folder and file name info
    # ------------------------------------------------------------------------------------------------------------------

    print('Data file: \n' + fileToAnalyse + '\n')

    dataDir = sep.join(fileToAnalyse.split(sep)[0:-3]) + sep
    flyID = fileToAnalyse.split(sep)[-2]
    expDir = sep.join(fileToAnalyse.split(sep)[0:-1]) + sep

    if not ('males' in expDir.split(sep)[-4] or 'females'in expDir.split(sep)[-4] or 'FlyOverDebugging'):
        print('You selected an invalid data directory.\n' +
              'Expected folder structure of the selected path is some/path/to/experimentName/flyGender/rawData/')
        exit(1)

    FODataFile = fileToAnalyse.split(sep)[-1]
    FODataFiles = [filepath.split(sep)[-1] for filepath in glob(expDir + '*.txt')]
    FODataFiles = sorted(FODataFiles)

    trial = FODataFiles.index(FODataFile) + 1

    # create analysis dir
    analysisDir = dataDir + 'analysis/'
    try:
        mkdir(analysisDir)
    except OSError:
        print('Analysis directory already exists.')

    dataFileParts = FODataFile.split('_')
    titleString = 'fly ' + flyID + ' in ' + dataFileParts[2] + ' of ' + dataFileParts[3] + ', trial' + str(trial)
    print('Analysing experiment...\n')
    print(titleString + '\n')

    if 'Invisible' in FODataFile or 'invisible' in FODataFile:
        objecttype = 'invisible'
    elif 'Empty' in FODataFile or 'empty' in FODataFile:
        objecttype = 'none'
    else:
        objecttype = 'visible'

    # ------------------------------------------------------------------------------------------------------------------
    # Load or read in logfile data
    # ------------------------------------------------------------------------------------------------------------------

    # Read in logfile data, parse header ...............................................................................
    header, FOData, numFrames, frameRange, calibParams, coordFile = loadSingleVRLogfile(expDir, FODataFile)

    # Read in object coordinates .......................................................................................
    visibleObjectCoords, invisibleObjectCoords, origin = loadObjectCoords(dataDir, coordFile)

    # Compute movement velocities ......................................................................................
    logTime = np.copy(FOData[:, 0])
    time = np.linspace(0, logTime[-1], numFrames)

    angle = convertRawHeadingAngle(FOData[:, 5])

    # N = 60
    # vTrans, vRot, vTransFilt, vRotFilt = velocityFromTrajectory(time, angle, FOData[:, 1], FOData[:, 2], N, numFrames)

    # Down sample data to 20 Hz ........................................................................................
    samplingRate = 20
    time_ds, xPos_ds, yPos_ds, angle_ds, numFrames_ds\
        = donwsampleFOData(samplingRate, logTime, time, FOData[:, 1], FOData[:, 2], angle)

    # and compute downsampled velocities
    N = 5
    vTrans_ds, vRot_ds, vTransFilt_ds, vRotFilt_ds\
        = velocityFromTrajectory(time_ds, angle_ds, xPos_ds, yPos_ds, N, numFrames_ds)

    # ------------------------------------------------------------------------------------------------------------------
    # Generate basic analysis plots
    # ------------------------------------------------------------------------------------------------------------------
    # Time step plot ...................................................................................................

    tstpfig = plt.figure(figsize=(10, 3))
    gs = gridspec.GridSpec(1, 2, width_ratios=np.hstack((2, 1)))
    tstpfig.suptitle(titleString, fontsize=14)
    histRange = (0, 0.011)

    ax = tstpfig.add_subplot(gs[0])
    ax.plot(FOData[0:-2, 0], (FOData[1:-1, 0]-FOData[0:-2, 0]).astype('float'), '.', alpha=0.1)
    ax.set_ylim(histRange)
    ax.set_xlim((0, time[-1]))
    ax.set_xlabel('time [s]')
    ax.set_ylabel('time step [1/s]')
    myAxisTheme(ax)

    ax = tstpfig.add_subplot(gs[1])
    ax.hist(FOData[1:-1, 0]-FOData[0:-2, 0], 50, histRange)
    ax.set_xlabel('time step [1/s]')
    ax.set_ylabel('count')
    ax.set_title('mean time step = ' + str(round(np.mean((FOData[1:-1, 0]-FOData[0:-2, 0])*1000.0), 2)) + 'ms')
    myAxisTheme(ax)

    tstpfig.tight_layout()

    makeNestedPlotDirectory(analysisDir, 'timeStepPlot/', objecttype + sep)

    tstpfig.savefig(analysisDir + 'timeStepPlot/' + objecttype + sep
                    + FODataFile[0:-4] + '_timeStepPlot_trial' + str(trial) + '.pdf', format='pdf')

    # Plot of walking trace (+ colorbar for time) with object locations ................................................
    tStart = 0
    tEnd = len(FOData[:, 1])
    tStep = 72
    frameRange = range(tStart, tEnd, tStep)
    colMap = 'Accent'
    arrowLength = 5

    trajfig = plt.figure(figsize=(10, 10))
    gs = gridspec.GridSpec(2, 1, height_ratios=np.hstack((10, 1)))

    axTraj = trajfig.add_subplot(gs[0])
    axTime = trajfig.add_subplot(gs[1])

    plotPosInRange(axTraj, axTime, frameRange, FOData[:, 0], FOData[:, 1], FOData[:, 2], np.pi/180*FOData[:, 5],
                   colMap, arrowLength, 0.5, 5)
    axTraj.scatter(visibleObjectCoords[:, 0], visibleObjectCoords[:, 1], 50, alpha=0.5,
                   facecolor='black', edgecolors='none')
    axTraj.scatter(invisibleObjectCoords[4:, 0], invisibleObjectCoords[4:, 1], 50, alpha=0.5,
                   facecolors='none', edgecolors='black')
    axTraj.set_xlabel(header[1], fontsize=12)
    axTraj.set_ylabel(header[2], fontsize=12)
    axTraj.set_title('Walking trace of ' + titleString, fontsize=14)
    axTraj.set_xlim([max(-650, min(FOData[:, 1]) - 20), min(650, max(FOData[:, 1]) + 20)])
    axTraj.set_ylim([max(-650, min(FOData[:, 2]) - 20), min(650, max(FOData[:, 2]) + 20)])
    myAxisTheme(axTraj)

    axTime.set_xlabel(header[0], fontsize=12)
    plt.xlim((0, FOData[-1, 0]))
    timeAxisTheme(axTime)

    makeNestedPlotDirectory(analysisDir, 'tracePlot/', objecttype + sep)

    trajfig.savefig(analysisDir + 'tracePlot/' + objecttype + sep
                    + FODataFile[0:-4] + '_traceObjectPlot_trial' + str(trial) + '.pdf', format='pdf')

    # Plot velocity distributions of downs sampled data ................................................................
    rotLim = (-5, 5)
    transLim = (0, 30)
    angleLim = (-np.pi, np.pi)
    summaryVeloFig_ds = velocitySummaryPlot(time_ds, vTrans_ds, vTransFilt_ds, vRot_ds, vRotFilt_ds, angle_ds, rotLim,
                                            transLim, angleLim, 'Down sampled velocity traces, ' + titleString)

    makeNestedPlotDirectory(analysisDir, 'velocityTraces/', objecttype + sep)

    summaryVeloFig_ds.savefig(analysisDir + 'velocityTraces/' + objecttype + sep
                              + FODataFile[0:-4] + '_veloTraces_ds_trial' + str(trial) + '.pdf', format='pdf')

    # ------------------------------------------------------------------------------------------------------------------
    #  Collapse traces to single object cell and plot resulting trace
    # ------------------------------------------------------------------------------------------------------------------

    # Collapse to 'mini-arena' while preserving the global heading .....................................................
    arenaRad = 60 # 1/2 distance between cones
    if objecttype == 'visible':
        objectCoords = np.copy(visibleObjectCoords[0:-3, 0:2])
    else:  # use of non-physics, invisible objects allows us to mark virtual object positions in empty arena
        objectCoords = np.copy(invisibleObjectCoords[4:, 0:2])

    xPosMA, yPosMA = collapseToMiniArena(FOData[:, 1], FOData[:, 2], arenaRad, objectCoords)

    # Compute donw sampled collapsed traces
    f_xPosMA = interp1d(time, xPosMA, kind='linear')
    f_yPosMA = interp1d(time, yPosMA, kind='linear')

    xPosMA_ds = f_xPosMA(time_ds)
    yPosMA_ds = f_yPosMA(time_ds)

    # Plot collapsed, down sampled trace ...............................................................................
    tStart = 0
    tEnd = numFrames_ds
    tStep = 4
    frameRange = range(tStart, tEnd, tStep)
    colMap = 'Accent'

    colTrFig = plt.figure(figsize=(9, 10))
    gs = gridspec.GridSpec(2, 1, height_ratios=np.hstack((10, 1)))

    colTrFig.suptitle('Collapsed walking trace ("mini arena" with central object)\n' + titleString, fontsize=14)

    axTraj = colTrFig.add_subplot(gs[0])
    axTime = colTrFig.add_subplot(gs[1])
    plotPosInRange(axTraj, axTime, frameRange, time_ds, xPosMA_ds, yPosMA_ds, angle_ds, colMap, 4, 0.5, 7)
    axTraj.plot(0, 0, marker='o', markersize=20, linestyle='none', alpha=0.5, color='black')
    axTraj.set_xlabel(header[1], fontsize=12)
    axTraj.set_ylabel(header[2], fontsize=12)
    axTraj.set_ylim([-(arenaRad+5), arenaRad + 5])
    axTraj.set_xlim([-(arenaRad+5), arenaRad + 5])
    myAxisTheme(axTraj)
    axTime.set_xlabel(header[0], fontsize=12)
    plt.xlim((0, time_ds[-1]))
    timeAxisTheme(axTime)

    makeNestedPlotDirectory(analysisDir, 'tracePlotMA/', objecttype + sep)

    colTrFig.savefig(analysisDir + 'tracePlotMA/' + objecttype + sep
                     + FODataFile[0:-4] + '_traceObjectPlot_ds_trial' + str(trial) + '.pdf', format='pdf')

    # ------------------------------------------------------------------------------------------------------------------
    # Compute heading angle relative to closest object (use collapsed coordinates)
    # ------------------------------------------------------------------------------------------------------------------

    # Compute parameters characterising fly's relationship to object
    objLocation = [0, 0]
    objDirection, objDistance, gammaFull, gamma, gammaV\
        = relationToObject(time_ds, xPosMA_ds, yPosMA_ds, angle_ds, objLocation)

    # Change in heading rel. to object .................................................................................
    sf = 0
    ef = len(xPosMA_ds)
    near = 6
    far = arenaRad
    vTransTH = 2

    turnTH = 3*np.std(abs(vRotFilt_ds))
    turnMask = (abs(vRotFilt_ds) > turnTH)

    selectedRangeDistAll = np.logical_and(objDistance > near, objDistance < far)

    selectedRangeDist = np.logical_and(np.logical_and(objDistance > near, objDistance < far), vTrans_ds > vTransTH)
    selectedRangeDistTurnWalk = np.logical_and(selectedRangeDist, turnMask)
    selectedRangeDistTurn = np.logical_and(np.logical_and(objDistance > near, objDistance < far), turnMask)

    headTurnFig = plt.figure(figsize=(15, 5))

    headTurnFig.suptitle('Relationship between turns and relativel heading angle\n' + flyID + ', trial' + str(trial) +
                         ', percentage turns: ' + str(round(100.0*sum(turnMask)/len(vRotFilt_ds), 2)) +
                         ' (within range ' + str(near) + '-' + str(far) + 'mm, trans. vel. > ' + str(vTransTH) + ')\n',
                         fontsize=13)

    distRange = (near, far)
    angleRange = (0, np.pi)
    vHeadRange = (-5, 5)

    ax0 = headTurnFig.add_subplot(131)
    ax0.plot(xPosMA_ds[sf:ef], yPosMA_ds[sf:ef], '.', color='grey', alpha=0.4)
    ax0.plot(xPosMA_ds[turnMask[sf:ef]], yPosMA_ds[turnMask[sf:ef]], '.', color='red', alpha=0.7)
    ax0.plot(0, 0, marker='o', markersize=15, linestyle='none', alpha=0.5, color='black')
    plt.xlabel('x [mm]')
    plt.ylabel('y [mm]')
    ax0.set_aspect('equal')
    myAxisTheme(ax0)

    ax1 = headTurnFig.add_subplot(132)
    niceScatterPlot(ax1, objDistance[selectedRangeDist], gamma[selectedRangeDist], distRange, angleRange, 'grey', 0.3)
    niceScatterPlot(ax1, objDistance[selectedRangeDistTurn], gamma[selectedRangeDistTurn], distRange, angleRange,
                    'red', 0.7)
    plt.xlabel('distance from object')
    plt.ylabel('heading angle')

    ax2 = headTurnFig.add_subplot(133)
    niceScatterPlot(ax2, gammaV[selectedRangeDist], gamma[selectedRangeDist], vHeadRange, angleRange, 'grey', 0.3)
    niceScatterPlot(ax2, gammaV[selectedRangeDistTurn], gamma[selectedRangeDistTurn], vHeadRange, angleRange,
                    'red', 0.7)
    plt.xlabel('change in heading angle (pos - increase, neg - decrease)')
    plt.ylabel('heading angle')

    headTurnFig.tight_layout()

    makeNestedPlotDirectory(analysisDir, 'relativeHeading/', objecttype + sep)

    headTurnFig.savefig(analysisDir + 'relativeHeading/' + objecttype + sep
                        + FODataFile[0:-4] + '_headingAndTurns_ds_trial' + str(trial) + '.pdf', format='pdf')

    # Visualise effect of turns ........................................................................................
    turnEffectFig = plt.figure(figsize=(12, 6))
    gs = gridspec.GridSpec(3, 3, height_ratios=np.hstack((1, 1.5, 1.5)), width_ratios=np.hstack((1.5, 1, 1)))

    ax0 = turnEffectFig.add_subplot(gs[:, 0])
    ax0.plot(xPosMA_ds[selectedRangeDist], yPosMA_ds[selectedRangeDist], '.', color='grey', alpha=0.4)
    ax0.plot(xPosMA_ds[selectedRangeDistTurn], yPosMA_ds[selectedRangeDistTurn], '.', color='red', alpha=0.7)
    ax0.plot(0, 0, marker='o', markersize=15, linestyle='none', alpha=0.5, color='black')
    plt.xlabel('x [mm]')
    plt.ylabel('y [mm]')
    ax0.set_xlim((-arenaRad, arenaRad))
    ax0.set_ylim((-arenaRad, arenaRad))
    ax0.set_aspect('equal')
    myAxisTheme(ax0)
    ax0.set_title('Effect of turns in ' + flyID + ', trial' + str(trial) +
                  '\n percentage turns: ' + str(round(100.0*sum(turnMask)/len(vRotFilt_ds), 2)) + '\n', fontsize=13)

    ax = turnEffectFig.add_subplot(gs[0, 1:3])
    gammaFullSelect = gammaFull[selectedRangeDistAll]
    gammaSelect = gamma[selectedRangeDistAll]
    try:
        plt.hist(gammaFullSelect[~np.isnan(gammaFullSelect)], bins=50, color='lightblue', alpha=0.8)
        plt.hist(gammaSelect[~np.isnan(gammaSelect)], bins=50, color='grey', alpha=0.5)
    except ValueError:
        print('Not enough values for histogram.')
    plt.xlabel('relative heading angle [rad], not filtered for vTrans > ' + str(vTransTH))
    plt.ylabel('count')
    ax.set_xlim((-np.pi, np.pi))
    myAxisTheme(ax)

    headingHist = plotVeloHistogram_fancy(gamma[selectedRangeDist], gs[1, 1], (0, np.pi), 'grey', 0.5)
    headingHist.set_ylabel('walking filtered\n count (vTrans > ' + str(vTransTH) + ')')

    headingHistTurn = plotVeloHistogram_fancy(gamma[selectedRangeDistTurn], gs[2, 1], (0, np.pi), 'red', 0.5)
    headingHistTurn.set_ylabel('walking & turn filtered\n count')
    headingHistTurn.set_xlabel('heading angle\n [rad]')

    rotVHist = plotVeloHistogram_fancy(gammaV[selectedRangeDist], gs[1, 2], (-10, 10), 'grey', 0.5)
    rotVHist.set_xlabel('change in heading while walking\n [rad/s]')

    rotVHistFilt = plotVeloHistogram_fancy(gammaV[selectedRangeDistTurn], gs[2, 2], (-10, 10), 'red', 0.5)
    rotVHistFilt.set_xlabel('change in heading during turn\n [rad/s]')

    turnEffectFig.tight_layout()

    makeNestedPlotDirectory(analysisDir, 'effectOfTurn/', objecttype + sep)

    turnEffectFig.savefig(analysisDir + 'effectOfTurn/' + objecttype + sep
                          + FODataFile[0:-4] + '_turnHeadingChange_trial' + str(trial) + '.pdf', format='pdf')

    # Directional modulation of runs (Gomez-Marin and Louis, 2014) .....................................................
    # turnModfig = plt.figure(figsize=(12, 7))
    # turnMod = modulationOfRuns(turnModfig, gammaFull, vRotFilt_ds,
    #                           selectedRangeDist, selectedRangeDistTurn, objDistance)
    #
    # turnMod.set_title(flyID + ' in VR (within range ' + str(near) + '-' + str(far) + 'mm)',  fontsize=13)
    #
    # makeNestedPlotDirectory(analysisDir, 'headingVsRotation/', objecttype + sep)
    #
    # turnModfig.savefig(analysisDir + 'headingVsRotation/' + objecttype + sep
    #                   + FODataFile[0:-4] + '_headingVsRotation_trial' + str(trial) + '.pdf', format='pdf')

    # ------------------------------------------------------------------------------------------------------------------
    # Save position and velocities for future analysis
    # ------------------------------------------------------------------------------------------------------------------

    toSave = {'time': time_ds,
              'xPos': xPos_ds,
              'yPos': yPos_ds,
              'xPosInMiniarena': xPosMA_ds,
              'yPosInMiniarena': yPosMA_ds,
              'headingAngle': angle_ds,
              'rotVelo': vRot_ds,
              'transVelo': vTrans_ds,
              'objectDistance': objDistance,
              'gammaFull': gammaFull,
              'gamma': gamma}
    # Alternatively use numpy array:
    # toSave=np.vstack((time_ds,xPos_ds, yPos_ds, xPosMA_ds, yPosMA_ds, angle_ds, vRot_ds,vTrans_ds,objDistance,gamma))

    # Save data in this format as *.npy for easy loading..
    np.save(expDir + FODataFile[:-4], toSave)

    # ------------------------------------------------------------------------------------------------------------------
    print('\n \n Analysis ran successfully. \n \n')
    # ------------------------------------------------------------------------------------------------------------------

    return 0
コード例 #2
0
def singleVROptogenTrialAnalysis(fileToAnalyse):
    # fileToAnalyse should be to complete path to the logfile of the FlyOver trial to be analysed.

    # TODO: add savePlots,recomputeFlyData as function arguments

    # ------------------------------------------------------------------------------------------------------------------
    # Extract folder and file name info
    # ------------------------------------------------------------------------------------------------------------------

    print('Data file: \n' + fileToAnalyse + '\n')

    dataDir = sep.join(fileToAnalyse.split(sep)[0:-3]) + sep
    flyID = fileToAnalyse.split(sep)[-2]
    expDir = sep.join(fileToAnalyse.split(sep)[0:-1]) + sep

    if not ('males' in expDir.split(sep)[-4] or 'females'in expDir.split(sep)[-4] or 'FlyOverDebugging'):
        print('You selected an invalid data directory.\n' +
              'Expected folder structure of the selected path is some/path/to/experimentName/flyGender/rawData/')
        exit(1)

    genotype = expDir.split(sep)[-5]

    # create analysis dir
    analysisDir = sep.join(dataDir.split(sep)[:-1]) + sep + 'analysis' + sep
    try:
        mkdir(analysisDir)
    except OSError:
        print('Analysis directory already exists.')

    FODataFile = fileToAnalyse.split(sep)[-1]
    FODataFiles = [filepath.split(sep)[-1] for filepath in glob(expDir + '*.txt')]
    FODataFiles = sorted(FODataFiles)

    trial = FODataFiles.index(FODataFile) + 1

    # ------------------------------------------------------------------------------------------------------------------
    # Load or read in logfile data
    # ------------------------------------------------------------------------------------------------------------------

    # Read in logfile data, parse header ...............................................................................
    header, FOData, numFrames, frameRange, calibParams, coordFile = loadSingleVRLogfile(expDir, FODataFile)

    if 'rZones' in coordFile:
        rZones = 'on'
    else:
        rZones = 'off'

    if 'invisible' in coordFile or 'Invisible' in coordFile:
        invisible = 'on'
        objecttype = 'invisible'

    else:
        invisible = 'off'
        objecttype = 'visible'

    dataFileParts = FODataFile.split('_')

    titleString = genotype + ' fly "' + flyID + '" in ' + dataFileParts[0] + ' of ' + dataFileParts[1] + 's\n' + \
        'with reinforcement zones ' + rZones + ', trial' + str(trial)
    print(titleString)

    # Extract reinforcement zone parameter .............................................................................
    rZone_rInner, rZone_rOuter, rZone_max, rZone_bl, rZone_gExp = rZoneParamsFromLogFile(expDir, FODataFile)

    # Read in object coordinates .......................................................................................
    visibleObjectCoords, invisibleObjectCoords, origin = loadObjectCoords(dataDir, coordFile)

    # Compute movement velocities ......................................................................................
    logTime = np.copy(FOData[:, 0])
    time = np.linspace(0, logTime[-1], numFrames)

    angle = convertRawHeadingAngle(FOData[:, 5])

    # N = 60
    # vTrans, vRot, vTransFilt, vRotFilt = velocityFromTrajectory(time, angle, FOData[:, 1], FOData[:, 2], N, numFrames)

    # Down sample data to 20 Hz ........................................................................................
    samplingRate = 20
    time_ds, xPos_ds, yPos_ds, angle_ds, numFrames_ds \
        = donwsampleFOData(samplingRate, logTime, time, FOData[:, 1], FOData[:, 2], angle)

    # and compute downsampled velocities
    N = 5
    vTrans_ds, vRot_ds, vTransFilt_ds, vRotFilt_ds \
        = velocityFromTrajectory(time_ds, angle_ds, xPos_ds, yPos_ds, N, numFrames_ds)

    # ------------------------------------------------------------------------------------------------------------------
    # Generate basic analysis plots
    # ------------------------------------------------------------------------------------------------------------------
    # Time step plot ...................................................................................................

    plotStp = 5
    tstpfig = plotFlyVRtimeStp(plotStp, FOData[:, 0], titleString)

    makeNestedPlotDirectory(analysisDir, 'timeStepPlot/', 'rZones_' + rZones + sep)

    tstpfig.savefig(analysisDir + 'timeStepPlot/' + 'rZones_' + rZones + sep
                    + FODataFile[0:-4] + '_timeStepPlot_trial' + str(trial) + '.pdf', format='pdf')

    # Plot of walking trace (+ colorbar for time) with object locations ................................................
    tStart = 0
    tEnd = len(FOData[:, 1])
    tStep = 72
    frameRange = range(tStart, tEnd, tStep)
    colMap = 'Accent'
    arrowLength = 5

    trajfig = plt.figure(figsize=(10, 10))
    gs = gridspec.GridSpec(2, 1, height_ratios=np.hstack((10, 1)))

    axTraj = trajfig.add_subplot(gs[0])
    axTime = trajfig.add_subplot(gs[1])

    plotPosInRange(axTraj, axTime, frameRange, FOData[:, 0], FOData[:, 1], FOData[:, 2], np.pi/180*FOData[:, 5],
                   colMap, arrowLength, 0.5, 5)
    axTraj.scatter(visibleObjectCoords[:, 0], visibleObjectCoords[:, 1], 50, alpha=0.5,
                   facecolor='black', edgecolors='none')
    axTraj.scatter(invisibleObjectCoords[4:, 0], invisibleObjectCoords[4:, 1], 50, alpha=0.5,
                   facecolors='none', edgecolors='black')
    axTraj.set_xlabel(header[1], fontsize=12)
    axTraj.set_ylabel(header[2], fontsize=12)
    axTraj.set_title('Walking trace of ' + titleString, fontsize=14)
    axTraj.set_xlim([max(-650, min(FOData[:, 1]) - 20), min(650, max(FOData[:, 1]) + 20)])
    axTraj.set_ylim([max(-650, min(FOData[:, 2]) - 20), min(650, max(FOData[:, 2]) + 20)])
    myAxisTheme(axTraj)

    axTime.set_xlabel(header[0], fontsize=12)
    plt.xlim((0, FOData[-1, 0]))
    timeAxisTheme(axTime)

    makeNestedPlotDirectory(analysisDir, 'tracePlot/', 'rZones_' + rZones + sep)

    trajfig.savefig(analysisDir + 'tracePlot/' + 'rZones_' + rZones + sep
                    + FODataFile[0:-4] + '_traceObjectPlot_trial' + str(trial) + '.pdf', format='pdf')

    # Visualise strength of optogenetic stimulation ....................................................................

    rEvents = FOData[:, 11]
    # downsample rEvents
    f_rEvents = interp1d(time, rEvents, kind='linear')
    rEvents_ds = f_rEvents(time_ds)

    tStep = 72
    frameRange = range(tStart, tEnd, tStep)

    trajRZfig = plt.figure(figsize=(10, 10))
    axTraj = trajRZfig.add_subplot(111)

    axTraj.scatter(visibleObjectCoords[:, 0], visibleObjectCoords[:, 1], 50, alpha=0.5,
                   facecolor='black', edgecolors='none')
    axTraj.scatter(invisibleObjectCoords[4:, 0], invisibleObjectCoords[4:, 1], 50, alpha=0.5,
                   facecolors='none', edgecolors='black')

    plt.plot(FOData[frameRange, 1], FOData[frameRange, 2], marker='.', markerfacecolor='grey',
             markeredgecolor='none', alpha=0.25)

    # overlay reinforcement value
    axTraj.scatter(FOData[frameRange, 1], FOData[frameRange, 2], s=rEvents[frameRange]*10,
                   c=rEvents[frameRange]*10, alpha=0.7, cmap=plt.get_cmap('Reds'))

    axTraj.set_xlabel(header[1], fontsize=12)
    axTraj.set_ylabel(header[2], fontsize=12)
    axTraj.set_title('Walking trace of ' + titleString, fontsize=14)
    axTraj.set_xlim([max(-650, min(FOData[:, 1]) - 20), min(650, max(FOData[:, 1]) + 20)])
    axTraj.set_ylim([max(-650, min(FOData[:, 2]) - 20), min(650, max(FOData[:, 2]) + 20)])
    axTraj.set_aspect('equal')
    myAxisTheme(axTraj)

    makeNestedPlotDirectory(analysisDir, 'tracePlotRZ/', 'rZones_' + rZones + sep)

    trajRZfig.savefig(analysisDir + 'tracePlotRZ/' + 'rZones_' + rZones + sep
                      + FODataFile[0:-4] + '_traceObjectPlot_trial' + str(trial) + '.pdf', format='pdf')

    # Plot velocity distributions of downs sampled data ................................................................
    rotLim = (-5, 5)
    transLim = (0, 30)
    angleLim = (-np.pi, np.pi)
    summaryVeloFig_ds = velocitySummaryPlot(time_ds, vTrans_ds, vTransFilt_ds, vRot_ds, vRotFilt_ds, angle_ds, rotLim,
                                            transLim, angleLim, 'Down sampled velocity traces, ' + titleString)

    makeNestedPlotDirectory(analysisDir, 'velocityTraces/', 'rZones_' + rZones + sep)

    summaryVeloFig_ds.savefig(analysisDir + 'velocityTraces/' + 'rZones_' + rZones + sep
                              + FODataFile[0:-4] + '_veloTraces_ds_trial' + str(trial) + '.pdf', format='pdf')

    # ------------------------------------------------------------------------------------------------------------------
    #  Collapse traces to single object cell and plot resulting trace
    # ------------------------------------------------------------------------------------------------------------------

    # Collapse to 'mini-arena' while preserving the global heading .....................................................
    arenaRad = 60 # 1/2 distance between cones
    if invisible == 'off':
        objectCoords = np.copy(visibleObjectCoords[0:-3, 0:2])
    else:  # use of non-physics, invisible objects allows us to mark virtual object positions in empty arena
        objectCoords = np.copy(invisibleObjectCoords[4:, 0:2])

    xPosMA, yPosMA = collapseToMiniArena(FOData[:, 1], FOData[:, 2], arenaRad, objectCoords)

    # Compute donw sampled collapsed traces
    f_xPosMA = interp1d(time, xPosMA, kind='linear')
    f_yPosMA = interp1d(time, yPosMA, kind='linear')

    xPosMA_ds = f_xPosMA(time_ds)
    yPosMA_ds = f_yPosMA(time_ds)

    # Plot collapsed, down sampled trace ...............................................................................
    tStart = 0
    tEnd = numFrames_ds
    tStep = 4
    frameRange = range(tStart, tEnd, tStep)
    colMap = 'Accent'

    colTrFig = plt.figure(figsize=(9, 10))
    gs = gridspec.GridSpec(2, 1, height_ratios=np.hstack((10, 1)))

    colTrFig.suptitle('Collapsed walking trace ("mini arena" with central object)\n' + titleString, fontsize=14)

    axTraj = colTrFig.add_subplot(gs[0])
    axTime = colTrFig.add_subplot(gs[1])
    plotPosInRange(axTraj, axTime, frameRange, time_ds, xPosMA_ds, yPosMA_ds, angle_ds, colMap, 4, 0.5, 7)

    if invisible == 'off':
        axTraj.plot(0, 0, marker='o', markersize=20, linestyle='none', alpha=0.75, color='black')
    else:
        axTraj.plot(0, 0, marker='o', markersize=20, alpha=0.75, markeredgewidth=0.5,
                    markerfacecolor='None', markeredgecolor='black')

    if rZones == 'on':
        rZoneRange = float(rZone_rOuter - rZone_rInner)
        for zRad in range(rZone_rInner, rZone_rOuter):
            circle1 = plt.Circle((0, 0), zRad, color='r', alpha=1.0/rZoneRange)
            axTraj.add_artist(circle1)

    axTraj.set_xlabel(header[1], fontsize=12)
    axTraj.set_ylabel(header[2], fontsize=12)
    axTraj.set_ylim([-(arenaRad+5), arenaRad + 5])
    axTraj.set_xlim([-(arenaRad+5), arenaRad + 5])
    myAxisTheme(axTraj)

    axTime.set_xlabel(header[0], fontsize=12)
    plt.xlim((0, time_ds[-1]))
    timeAxisTheme(axTime)

    makeNestedPlotDirectory(analysisDir, 'tracePlotMA/', 'rZones_' + rZones + sep)

    colTrFig.savefig(analysisDir + 'tracePlotMA/' + 'rZones_' + rZones + sep
                     + FODataFile[0:-4] + '_traceObjectPlot_ds_trial' + str(trial) + '.pdf', format='pdf')

    # ------------------------------------------------------------------------------------------------------------------
    # Compute heading angle relative to closest object (use collapsed coordinates)
    # ------------------------------------------------------------------------------------------------------------------

    # Compute parameters characterising fly's relationship to object ...................................................
    objLocation = [0, 0]
    objDirection, objDistance, gammaFull, gamma, gammaV \
        = relationToObject(time_ds, xPosMA_ds, yPosMA_ds, angle_ds, objLocation)

    # Change in heading rel. to object .................................................................................
    near = 6
    far = arenaRad
    vTransTH = 2

    turnTH = 2.5*np.std(abs(vRotFilt_ds))
    turnMask = (abs(vRotFilt_ds) > turnTH)

    preTurnMask = np.hstack((turnMask[samplingRate/20:], np.zeros(samplingRate/20)))

    selectedRangeDistAll = np.logical_and(objDistance > near, objDistance < far)

    selectedRangeDist = np.logical_and(np.logical_and(objDistance > near, objDistance < far), vTrans_ds > vTransTH)
    selectedRangeDistPreTurnWalk = np.logical_and(selectedRangeDist, preTurnMask)
    selectedRangeDistPreTurn = np.logical_and(np.logical_and(objDistance > near, objDistance < far), preTurnMask)
    selectedRangeDistTurn = np.logical_and(np.logical_and(objDistance > near, objDistance < far), turnMask)

    # Visualise effect of turns ........................................................................................
    turnEffectFig = plt.figure(figsize=(12, 6))
    gs = gridspec.GridSpec(3, 3, height_ratios=np.hstack((1, 1.5, 1.5)), width_ratios=np.hstack((1.5, 1, 1)))

    ax0 = turnEffectFig.add_subplot(gs[:, 0])
    ax0.plot(xPosMA_ds[selectedRangeDist], yPosMA_ds[selectedRangeDist], '.', color='grey', alpha=0.4)
    ax0.plot(xPosMA_ds[selectedRangeDistTurn], yPosMA_ds[selectedRangeDistTurn], '.', color='lightblue', alpha=0.7)
    ax0.plot(xPosMA_ds[selectedRangeDistPreTurn], yPosMA_ds[selectedRangeDistPreTurn], '.', color='red', alpha=0.4)
    ax0.plot(0, 0, marker='o', markersize=15, linestyle='none', alpha=0.5, color='black')
    plt.xlabel('x [mm]')
    plt.ylabel('y [mm]')
    ax0.set_xlim((-arenaRad, arenaRad))
    ax0.set_ylim((-arenaRad, arenaRad))
    ax0.set_aspect('equal')
    myAxisTheme(ax0)
    ax0.set_title('Effect of turns in ' + flyID + ', trial' + str(trial) +
                  '\n percentage turns: ' + str(round(100.0*sum(turnMask)/len(vRotFilt_ds), 2)) + '\n', fontsize=13)

    ax = turnEffectFig.add_subplot(gs[0, 1:3])
    gammaFullSelect = gammaFull[selectedRangeDistAll]
    gammaSelect = gamma[selectedRangeDistAll]
    try:
        plt.hist(gammaFullSelect[~np.isnan(gammaFullSelect)], bins=50, color='lightblue', alpha=0.8)
        plt.hist(gammaSelect[~np.isnan(gammaSelect)], bins=50, color='grey', alpha=0.5)
    except ValueError:
        print('Not enough values for histogram.')
    plt.xlabel('relative heading angle [rad], not filtered for vTrans > ' + str(vTransTH))
    plt.ylabel('count')
    ax.set_xlim((-np.pi, np.pi))
    myAxisTheme(ax)

    headingHist = plotVeloHistogram_fancy(gamma[selectedRangeDist], gs[1, 1], (0, np.pi), 'grey', 0.5)
    headingHist.set_ylabel('walking filtered\n count (vTrans > ' + str(vTransTH) + ')')

    headingHistTurn = plotVeloHistogram_fancy(gamma[selectedRangeDistPreTurnWalk], gs[2, 1], (0, np.pi), 'red', 0.5)
    headingHistTurn.set_ylabel('walking & turn filtered\n count')
    headingHistTurn.set_xlabel('heading angle\n [rad]')

    rotVHist = plotVeloHistogram_fancy(gammaV[selectedRangeDist], gs[1, 2], (-10, 10), 'grey', 0.5)
    rotVHist.set_xlabel('change in heading while walking\n [rad/s]')

    rotVHistFilt = plotVeloHistogram_fancy(gammaV[selectedRangeDistPreTurnWalk], gs[2, 2], (-10, 10), 'red', 0.5)
    rotVHistFilt.set_xlabel('change in heading during turn\n [rad/s]')

    turnEffectFig.tight_layout()

    makeNestedPlotDirectory(analysisDir, 'effectOfTurn/', 'rZones_' + rZones + sep)

    turnEffectFig.savefig(analysisDir + 'effectOfTurn/' + 'rZones_' + rZones + sep
                          + FODataFile[0:-4] + '_turnHeadingChange_trial' + str(trial) + '.pdf', format='pdf')

    # ------------------------------------------------------------------------------------------------------------------
    # Convert trajectory to polar coordinates and visualise trace and effect of turns
    # ------------------------------------------------------------------------------------------------------------------

    # transform trajectory to polar coordinates
    objDist, theta = cartesian2polar(xPosMA_ds, yPosMA_ds)

    # compute curvature
    polarCurv, d_theta, dtheta_objDist = polarCurvature(theta, objDist)

    d_objDist = np.hstack((0, np.diff(objDist)))

    # Compute sign of turn relative to object
    turnSign = np.sign(polarCurv)
    turnSign[d_theta > 0] = np.sign(polarCurv[d_theta > 0])
    turnSign[d_theta < 0] = -np.sign(polarCurv[d_theta < 0])

    # Compute curvature-based criterion for turns
    # q2, q98 = np.percentile(polarCurv[~np.isnan(polarCurv)], [2, 98])
    # curvSelect = abs(polarCurv) < (q98 - q2)/2
    # curvTurnMask_L = polarCurv > q98
    # curvTurnMask_R = polarCurv < q2

    # Generate filtered curvature for plots and curvature magnitude
    # polarCurvPlt = polarCurv[curvSelect]
    # curvMag = abs(polarCurv)
    # correctedPolarCurv = abs(polarCurv)*turnSign

    selectPts_apr = d_objDist < 0
    selectPts_dep = d_objDist > 0

    selectPts_apr_turnR = np.logical_and(selectPts_apr, vRotFilt_ds < -1*turnTH)
    selectPts_apr_turnL = np.logical_and(selectPts_apr, vRotFilt_ds > 1*turnTH)
    selectPts_dep_turnR = np.logical_and(selectPts_dep, vRotFilt_ds < -1*turnTH)
    selectPts_dep_turnL = np.logical_and(selectPts_dep, vRotFilt_ds > 1*turnTH)

    fig = plt.figure(figsize=(15, 10))

    xlimRange = (5, 60)

    ax = fig.add_subplot(211)
    ax = plotPolarTrace(ax, titleString + '\nApproaches to ' + objecttype + ' object',
                        selectPts_apr, selectPts_apr_turnR, selectPts_apr_turnL, objDist, gammaFull, vRot_ds, xlimRange)

    ax = fig.add_subplot(212)
    ax = plotPolarTrace(ax, 'Departures from ' + objecttype + ' object',
                        selectPts_dep, selectPts_dep_turnR, selectPts_dep_turnL, objDist, gammaFull, vRot_ds, xlimRange)

    makeNestedPlotDirectory(analysisDir, 'polarTrace/', 'rZones_' + rZones + sep)

    fig.savefig(analysisDir + 'polarTrace/' + 'rZones_' + rZones + sep
                + FODataFile[0:-4] + '_polarTrace_trial' + str(trial) + '.pdf', format='pdf')

    # ------------------------------------------------------------------------------------------------------------------
    # Save position and velocities for future analysis
    # ------------------------------------------------------------------------------------------------------------------

    toSave = {'time': time_ds,
              'xPos': xPos_ds,
              'yPos': yPos_ds,
              'xPosInMiniarena': xPosMA_ds,
              'yPosInMiniarena': yPosMA_ds,
              'headingAngle': angle_ds,
              'rotVelo': vRot_ds,
              'transVelo': vTrans_ds,
              'objectDistance': objDistance,
              'gammaFull': gammaFull,
              'gamma': gamma,
              'rEvents': rEvents_ds}

    # Alternatively use numpy array:
    # toSave=np.vstack((time_ds,xPos_ds, yPos_ds, xPosMA_ds, yPosMA_ds, angle_ds, vRot_ds,vTrans_ds,objDistance,gamma))

    # Save data in this format as *.npy for easy loading..
    np.save(expDir + FODataFile[:-4], toSave)

    # ------------------------------------------------------------------------------------------------------------------
    print('\n \n Analysis ran successfully. \n \n')
    # ------------------------------------------------------------------------------------------------------------------

    plt.close('all')

    return 0
コード例 #3
0
def singleTwoObjVROptogenTrialAnalysis(fileToAnalyse):
    # fileToAnalyse should be to complete path to the logfile of the FlyOver trial to be analysed.

    # TODO: add savePlots,recomputeFlyData as function arguments

    # ------------------------------------------------------------------------------------------------------------------
    # Extract folder and file name info
    # ------------------------------------------------------------------------------------------------------------------

    print('Data file: \n' + fileToAnalyse + '\n')

    dataDir = sep.join(fileToAnalyse.split(sep)[0:-3]) + sep
    flyID = fileToAnalyse.split(sep)[-2]
    expDir = sep.join(fileToAnalyse.split(sep)[0:-1]) + sep

    if not ('males' in expDir.split(sep)[-4] or 'females'in expDir.split(sep)[-4] or 'FlyOverDebugging'):
        print('You selected an invalid data directory.\n' +
              'Expected folder structure of the selected path is some/path/to/experimentName/flyGender/rawData/')
        exit(1)

    genotype = expDir.split(sep)[-5]

    # create analysis dir
    analysisDir = sep.join(dataDir.split(sep)[:-1]) + sep + 'analysis' + sep
    try:
        mkdir(analysisDir)
    except OSError:
        print('Analysis directory already exists.')

    FODataFile = fileToAnalyse.split(sep)[-1]
    FODataFiles = [filepath.split(sep)[-1] for filepath in glob(expDir + '*.txt')]
    FODataFiles = sorted(FODataFiles)

    trial = FODataFiles.index(FODataFile) + 1

    if ('train' in FODataFile):
        rZones = 'on'
    else:
        rZones = 'off'

    dataFileParts = FODataFile.split('_')

    trialType = dataFileParts[-3]
    invisible = 'off'
    objecttype = 'visible'

    titleString = genotype + ' fly "' + flyID + '" in ' + dataFileParts[0] + ' of ' + dataFileParts[1] + ' and ' + \
        dataFileParts[2] + '\n' + trialType + ' trial'

    print(titleString)

    # ------------------------------------------------------------------------------------------------------------------
    # Load or read in logfile data
    # ------------------------------------------------------------------------------------------------------------------

    # Read in logfile data, parse header ...............................................................................
    header, FOData, numFrames, frameRange, calibParams, coordFile = loadSingleVRLogfile(expDir, FODataFile)

    # Extract reinforcement zone parameter .............................................................................
    #rZone_rInner, rZone_rOuter, rZone_max, rZone_bl, rZone_gExp = rZoneParamsFromLogFile(expDir, FODataFile)

    # Read in object coordinates .......................................................................................
    visibleObjectCoords, visibleObjectName, invisibleObjectCoords, origin = loadObjectCoordIdentities(dataDir, coordFile)

    # Compute movement velocities ......................................................................................
    logTime = np.copy(FOData[:, 0])
    time = np.linspace(0, logTime[-1], numFrames)

    angle = convertRawHeadingAngle(FOData[:, 5])

    # N = 60
    # vTrans, vRot, vTransFilt, vRotFilt = velocityFromTrajectory(time, angle, FOData[:, 1], FOData[:, 2], N, numFrames)

    # Down sample data to 20 Hz ........................................................................................
    samplingRate = 20
    time_ds, xPos_ds, yPos_ds, angle_ds, numFrames_ds \
        = donwsampleFOData(samplingRate, logTime, time, FOData[:, 1], FOData[:, 2], angle)

    # and compute downsampled velocities
    N = 5
    vTrans_ds, vRot_ds, vTransFilt_ds, vRotFilt_ds \
        = velocityFromTrajectory(time_ds, angle_ds, xPos_ds, yPos_ds, N, numFrames_ds)

    # ------------------------------------------------------------------------------------------------------------------
    # Generate basic analysis plots
    # ------------------------------------------------------------------------------------------------------------------
    # Time step plot ...................................................................................................

    plotStp = 5
    tstpfig = plotFlyVRtimeStp(plotStp, FOData[:, 0], titleString)

    makeNestedPlotDirectory(analysisDir, 'timeStepPlot/', trialType + 'Trial' + sep)

    tstpfig.savefig(analysisDir + 'timeStepPlot/' + trialType + 'Trial' + sep + FODataFile[0:-4] + '_timeStepPlot.pdf',
                    format='pdf')

    # Plot of walking trace (+ colorbar for time) with object locations ................................................

    coneShape = np.asarray([bool('Cone' in objName) for objName in visibleObjectName])
    cyliShape = np.asarray([bool('Cyli' in objName) for objName in visibleObjectName])

    tStart = 0
    tEnd = len(FOData[:, 1])
    tStep = 72
    frameRange = range(tStart, tEnd, tStep)
    colMap = 'Accent'
    arrowLength = 4

    trajfig = plt.figure(figsize=(10, 10))
    gs = gridspec.GridSpec(2, 1, height_ratios=np.hstack((10, 1)))

    axTraj = trajfig.add_subplot(gs[0])
    axTime = trajfig.add_subplot(gs[1])

    plotPosInRange(axTraj, axTime, frameRange, FOData[:, 0], FOData[:, 1], FOData[:, 2], np.pi/180*FOData[:, 5],
                   colMap, arrowLength, 0.5, 5)
    axTraj.scatter(visibleObjectCoords[coneShape, 0], visibleObjectCoords[coneShape, 1], 50, alpha=0.5,
                   facecolors='black', edgecolors='none', marker='^')
    axTraj.scatter(visibleObjectCoords[cyliShape, 0], visibleObjectCoords[cyliShape, 1], 50, alpha=0.5,
                   facecolors='black', edgecolors='none', marker='s')
    #axTraj.scatter(invisibleObjectCoords[4:, 0], invisibleObjectCoords[4:, 1], 50, alpha=0.5,
    #               facecolors='none', edgecolors='black')
    axTraj.set_xlabel(header[1], fontsize=12)
    axTraj.set_ylabel(header[2], fontsize=12)
    axTraj.set_title('Walking trace of ' + titleString, fontsize=14)
    axTraj.set_xlim([max(-650, min(FOData[:, 1]) - 20), min(650, max(FOData[:, 1]) + 20)])
    axTraj.set_ylim([max(-650, min(FOData[:, 2]) - 20), min(650, max(FOData[:, 2]) + 20)])
    myAxisTheme(axTraj)

    axTime.set_xlabel(header[0], fontsize=12)
    plt.xlim((0, FOData[-1, 0]))
    timeAxisTheme(axTime)

    makeNestedPlotDirectory(analysisDir, 'tracePlot/', trialType + 'Trial' + sep)

    trajfig.savefig(analysisDir + 'tracePlot/' + trialType + 'Trial' + sep
                    + FODataFile[0:-4] + '_traceObjectPlot.pdf', format='pdf')

    # Visualise strength of optogenetic stimulation ....................................................................

    rEvents = FOData[:, 11]
    # downsample rEvents
    f_rEvents = interp1d(time, rEvents, kind='linear')
    rEvents_ds = f_rEvents(time_ds)

    tStep = 72
    frameRange = range(tStart, tEnd, tStep)

    trajRZfig = plt.figure(figsize=(10, 10))
    axTraj = trajRZfig.add_subplot(111)

    axTraj.scatter(visibleObjectCoords[coneShape, 0], visibleObjectCoords[coneShape, 1], 50, alpha=0.5,
                   facecolors='black', edgecolors='none', marker='^')
    axTraj.scatter(visibleObjectCoords[cyliShape, 0], visibleObjectCoords[cyliShape, 1], 50, alpha=0.5,
                   facecolors='black', edgecolors='none', marker='s')

    plt.plot(FOData[frameRange, 1], FOData[frameRange, 2], marker='.', markerfacecolor='grey',
             markeredgecolor='none', linestyle='none', alpha=0.1)

    #overlay reinforcement val
    axTraj.scatter(FOData[frameRange, 1], FOData[frameRange, 2], s=rEvents[frameRange]*3,
                   c=rEvents[frameRange]*3, alpha=0.7, cmap=plt.get_cmap('Reds'))

    axTraj.set_xlabel(header[1], fontsize=12)
    axTraj.set_ylabel(header[2], fontsize=12)
    axTraj.set_title('Walking trace of ' + titleString, fontsize=14)
    axTraj.set_xlim([max(-650, min(FOData[:, 1]) - 20), min(650, max(FOData[:, 1]) + 20)])
    axTraj.set_ylim([max(-650, min(FOData[:, 2]) - 20), min(650, max(FOData[:, 2]) + 20)])
    axTraj.set_aspect('equal')
    myAxisTheme(axTraj)

    makeNestedPlotDirectory(analysisDir, 'tracePlotRZ/', trialType + 'Trial' + sep)

    trajRZfig.savefig(analysisDir + 'tracePlotRZ/' + trialType + 'Trial' + sep
                    + FODataFile[0:-4] + '_traceObjectPlot.pdf', format='pdf')

    # Plot velocity distributions of downs sampled data ................................................................
    rotLim = (-5, 5)
    transLim = (0, 30)
    angleLim = (-np.pi, np.pi)
    summaryVeloFig_ds = velocitySummaryPlot(time_ds, vTrans_ds, vTransFilt_ds, vRot_ds, vRotFilt_ds, angle_ds, rotLim,
                                            transLim, angleLim, 'Down sampled velocity traces, ' + titleString)

    makeNestedPlotDirectory(analysisDir, 'velocityTraces/', trialType + 'Trial' + sep)

    summaryVeloFig_ds.savefig(analysisDir + 'velocityTraces/' + trialType + 'Trial' + sep
                              + FODataFile[0:-4] + '_veloTraces_ds.pdf', format='pdf')

    # ------------------------------------------------------------------------------------------------------------------
    #  Collapse traces to single object cell and plot resulting trace
    # ------------------------------------------------------------------------------------------------------------------

    # Collapse to 'mini-arena' while preserving the global heading .....................................................
    gridSize = 60.0 # closest distance between landmarks
    gridRepeat = (6, 5) # grid height in repeats of gridSize in x and y
    xPosMA, yPosMA = collapseTwoObjGrid(FOData[:, 1], FOData[:, 2], gridSize, gridRepeat)

    # Compute donw sampled collapsed traces
    f_xPosMA = interp1d(time, xPosMA, kind='linear')
    f_yPosMA = interp1d(time, yPosMA, kind='linear')

    xPosMA_ds = f_xPosMA(time_ds)
    yPosMA_ds = f_yPosMA(time_ds)

    # Plot collapsed, down sampled trace ...............................................................................
    tStart = 0
    tEnd = numFrames_ds
    tStep = 4
    frameRange = range(tStart, tEnd, tStep)
    colMap = 'Accent'

    colTrFig = plt.figure(figsize=(9, 10))
    gs = gridspec.GridSpec(2, 1, height_ratios=np.hstack((10, 1)))

    colTrFig.suptitle('Collapsed walking trace ("mini arena" with central object)\n' + titleString, fontsize=14)

    axTraj = colTrFig.add_subplot(gs[0])
    axTime = colTrFig.add_subplot(gs[1])
    plotPosInRange(axTraj, axTime, frameRange, time_ds, xPosMA_ds, yPosMA_ds, angle_ds, colMap, 4, 0.5, 7)

    axTraj.plot(gridSize/2, -gridSize/2, 'ks', markersize=8)
    axTraj.plot(gridSize/2, gridSize/2, 'k^', markersize=10)
    axTraj.plot(3*gridSize/2, -gridSize/2, 'k^', markersize=10)
    axTraj.plot(3*gridSize/2, gridSize/2, 'ks', markersize=8)

    axTraj.set_xlabel(header[1], fontsize=12)
    axTraj.set_ylabel(header[2], fontsize=12)
    axTraj.set_ylim([-(5+gridSize), gridSize+5])
    axTraj.set_xlim([-5, 2*gridSize+5])
    myAxisTheme(axTraj)

    axTime.set_xlabel(header[0], fontsize=12)
    plt.xlim((0, time_ds[-1]))
    timeAxisTheme(axTime)

    # if rZones == 'on':
    #    rZoneRange = float(rZone_rOuter - rZone_rInner)
    #    for zRad in range(rZone_rInner, rZone_rOuter):
    #        circle1 = plt.Circle((0, 0), zRad, color='r', alpha=1.0/rZoneRange)
    #        axTraj.add_artist(circle1)

    makeNestedPlotDirectory(analysisDir, 'tracePlotMA/', trialType + 'Trial' + sep)

    colTrFig.savefig(analysisDir + 'tracePlotMA/' + trialType + 'Trial' + sep
                     + FODataFile[0:-4] + '_traceObjectPlot_ds.pdf', format='pdf')

    # ------------------------------------------------------------------------------------------------------------------
    # Save position and velocities for future analysis
    # ------------------------------------------------------------------------------------------------------------------

    toSave = {'time': time_ds,
              'xPos': xPos_ds,
              'yPos': yPos_ds,
              'xPosInMiniarena': xPosMA_ds,
              'yPosInMiniarena': yPosMA_ds,
              'headingAngle': angle_ds,
              'rotVelo': vRot_ds,
              'transVelo': vTrans_ds,
              'rEvents': rEvents_ds}

    # Save data in this format as *.npy for easy loading..
    np.save(expDir + FODataFile[:-4], toSave)

    # ------------------------------------------------------------------------------------------------------------------
    print('\n \n Analysis ran successfully. \n \n')
    # ------------------------------------------------------------------------------------------------------------------

    plt.close('all')

    return 0
コード例 #4
0
def singleVRStripeTrialAnalysis(fileToAnalyse):

    datadir = sep.join(fileToAnalyse.split(sep)[0:-3]) + sep
    flyID = fileToAnalyse.split(sep)[-2]
    expDir = sep.join(fileToAnalyse.split(sep)[0:-1]) + sep

    FODataFile = fileToAnalyse.split(sep)[-1]

    print('\n')
    print('flyID: '+flyID+'\n')
    # # Load raw data

    dataFileParts = FODataFile.split('_')
    genotype = dataFileParts[2]
    sceneName = dataFileParts[0]
    titleString = 'fly '+flyID+' ('+genotype+')'+' in '+sceneName+' world'

    if 'Stripe' in sceneName or 'stripe' in sceneName:
        sceneType = 'stripe'
    else:
        sceneType = 'plane'

    print(titleString+'\n')
    print('scene type: ' + sceneType)

    # # Load data
    # ## Load FlyOver log file, extract calibration paramater and name of cood file
    header, FOData, numFrames, frameRange, calibParams, coordFile = loadSingleVRLogfile(expDir, FODataFile)

    # ## Read in object coordinates
    visObjCoords, visObjName, invisObjCoords, origin = loadObjectCoordIdentities(datadir, coordFile)
    print('Coord file: '+coordFile+'\n')

    # # Compute derrived values
    # ## Compute movement velocities
    logTime = np.copy(FOData[:, 0])
    time = np.linspace(0, logTime[-1], numFrames)

    angle = convertRawHeadingAngle(FOData[:, 5])

    # ## Collapse to 'mini-arena' while preserving the global heading  ---  if trialtype is 'plane'
    if sceneType == 'plane':
        xPos = FOData[:, 1]
        yPos = FOData[:, 2]
        arenaRad = 60 # 1/2 distance between objects
        objectCoords = np.copy(visObjCoords[0:-3, 0:2])

        xPosMA, yPosMA = collapseToMiniArena(xPos, yPos, arenaRad, objectCoords)

    # ## Compute fly's walking velocieties from TM raw values  ---  if trial type is 'stripe'
    if sceneType == 'stripe':
        dx1 = FOData[:, 6]
        dy1 = FOData[:, 7]
        dx2 = FOData[:, 8]
        dy2 = FOData[:, 9]

        dtime = np.diff(time)
        dtime = np.hstack((dtime[0], dtime))

        # parameter definitions
        gammaRad = 45*np.pi/180 # absolute angle of cameras to longitudinal axis (of fly)
        rBall = float(calibParams[0])
        pixel2mm = 0.013514

        conversionFactor_pitch = (1.0/pixel2mm)*float(calibParams[0])*2.0*(np.pi/2.0)*(1.0/float(calibParams[1]))
        conversionFactor_yaw = (1.0/pixel2mm)*float(calibParams[0])*2.0*(np.pi/2.0)*(1.0/float(calibParams[2]))

        # compute virtual rotation of fly
        vFwdBall = - (dy1 + dy2) * np.cos(gammaRad)
        vSideBall = - (dy1 - dy2) * np.sin(gammaRad)
        vRotBall = - (dx1 + dx2)/2

        # convert A.U. --> pixel --> mm
        vFwdBall = pixel2mm * vFwdBall * conversionFactor_pitch
        vSideBall = pixel2mm * vSideBall * ((conversionFactor_yaw + conversionFactor_pitch)/2)
        vRotBall = pixel2mm * vRotBall * conversionFactor_yaw

        # convert to mm/s
        vFwdBall = vFwdBall / dtime
        vSideBall = vSideBall / dtime
        vRotBall = vRotBall / dtime
        #  mm/s to deg/s
        vRot = -vRotBall / rBall

        # Assume initial position (0 0 0) = (x-coord, y-coord, angle):
        # --> fly in origin, aligned with x axis (head forward)
        # During measurement coordinate system is fly-centered, moves with fly. Compute all changes along those axes by
        # updating angle and projecting the position changes onto the fixed coordinate system
        angle = np.cumsum(vRot * dtime)
        angle = np.mod((angle + np.pi), 2*np.pi) - np.pi

        # movement in x and y direction
        yTM_i = vSideBall * np.cos(-angle) - vFwdBall * np.sin(-angle)
        yPos = np.cumsum(yTM_i * dtime)

        xTM_i = vSideBall * np.sin(-angle) + vFwdBall * np.cos(-angle)
        xPos = np.cumsum(xTM_i * dtime)

        vTrans = np.hypot(xTM_i, yTM_i)

    # ## Downsample
    # Down sample data to 20 Hz ........................................................................................
    samplingRate = 20
    time_ds, xPos_ds, yPos_ds, angle_ds, numFrames_ds = donwsampleFOData(samplingRate, logTime, time, xPos, yPos, angle)

    # and compute downsampled velocities
    N = 5
    vTrans_ds, vRot_ds, vTransFilt_ds, vRotFilt_ds \
        = velocityFromTrajectory(time_ds, angle_ds, xPos_ds, yPos_ds, N, numFrames_ds)

    if sceneType == 'plane':
        # Compute donwsampled collapsed traces
        f_xPosMA = interp1d(time, xPosMA, kind='linear')
        f_yPosMA = interp1d(time, yPosMA, kind='linear')

        xPosMA_ds = f_xPosMA(time_ds)
        yPosMA_ds = f_yPosMA(time_ds)

    # # Generate basic analysis plots
    # ## Time step plot
    tstpfig = plt.figure(figsize=(10, 3))
    gs = gridspec.GridSpec(1, 2, width_ratios=np.hstack((2, 1)))
    tstpfig.suptitle(titleString, fontsize=13)

    histRange = (0, 0.011)

    ax = tstpfig.add_subplot(gs[0])
    ax.plot(FOData[0:-2, 0], (FOData[1:-1, 0]-FOData[0:-2, 0]).astype('float'), '.', alpha=0.1)
    ax.set_ylim(histRange)
    ax.set_xlim((0, time[-1]))
    ax.set_xlabel('time [s]')
    ax.set_ylabel('time step [1/s]')
    myAxisTheme(ax)

    ax = tstpfig.add_subplot(gs[1])
    ax.hist(FOData[1:-1, 0]-FOData[0:-2, 0], 50, histRange)
    ax.set_xlabel('time step [1/s]')
    ax.set_ylabel('count')
    ax.set_title('mean time step = ' + str(round(np.mean((FOData[1:-1, 0]-FOData[0:-2, 0])*1000.0), 2)) + 'ms')
    myAxisTheme(ax)

    tstpfig.tight_layout()

    try:
        mkdir(datadir + 'analysis'+sep+'timeStepPlot'+sep)
    except OSError:
        print('Plot directory already exists.')

    tstpfig.savefig(datadir+'analysis'+sep+'timeStepPlot'+sep+FODataFile[0:-4]+'_timeStepPlot.pdf', format='pdf')

    # ## Plot velocity distributions of downsampled data
    rotLim = (-5, 5)
    transLim = (0, 30)
    angleLim = (-np.pi, np.pi)
    summaryVeloFig_ds = velocitySummaryPlot(time_ds, vTrans_ds, vTransFilt_ds, vRot_ds, vRotFilt_ds, angle_ds,
                                            rotLim, transLim, angleLim,
                                            'Downsampled and filtered downsampled velocity traces, ' + titleString)

    try:
        mkdir(datadir+'analysis'+sep+'velocityTraces'+sep)
    except OSError:
        print('Plot directory already exists.')

    summaryVeloFig_ds.savefig(datadir+'analysis'+sep+'velocityTraces'+sep+FODataFile[0:-4]+'_veloTraces_ds.pdf',
                              format='pdf')

    # ## Classify time point as 'moving' or 'non-moving' based on transl. velocity
    vTransTH = 2.0
    moving = vTrans_ds > vTransTH

    # ## Compute relative heading
    if sceneType == 'plane':
        objLocation = [0, 0]
        objDirection, objDistance, gammaFull, gamma, gammaV \
            = relationToObject(time_ds, xPosMA_ds, yPosMA_ds, angle_ds, objLocation)
    else: # sceneType == 'stripe'
        gamma = abs(angle_ds)
        gammaFull = angle_ds

    # ## Heading angle distribution (if stripe)
    headingfig = plt.figure(figsize=(10, 8))

    gammaPlt = headingfig.add_subplot(221)
    histRange = (0, np.pi)
    nhead, edges = np.histogram(gamma[moving > 0], density=True, range=histRange, bins=18)
    gammaPlt.plot(edges[:-1]+np.diff(edges)/2, nhead)
    gammaPlt.set_xlim(histRange)
    gammaPlt.set_xlabel('rel. heading')
    gammaPlt.set_ylabel('frequency (when moving)')
    myAxisTheme(gammaPlt)

    gammaFullPlt = headingfig.add_subplot(222)
    histRange = (-np.pi, np.pi)
    nhead, edges = np.histogram(gammaFull[moving > 0], density=True, range=histRange, bins=36)
    gammaFullPlt.plot(edges[:-1]+np.diff(edges)/2, nhead)
    gammaFullPlt.set_xlim(histRange)
    gammaFullPlt.set_xlabel('rel. heading (full)')
    myAxisTheme(gammaFullPlt)

    gammaPlt = headingfig.add_subplot(223)
    histRange = (0, np.pi)
    nhead, edges = np.histogram(gamma, density=True, range=histRange, bins=18)
    gammaPlt.plot(edges[:-1]+np.diff(edges)/2, nhead, color='grey')
    gammaPlt.set_xlim(histRange)
    gammaPlt.set_xlabel('rel. heading')
    gammaPlt.set_ylabel('frequency')
    myAxisTheme(gammaPlt)

    gammaFullPlt = headingfig.add_subplot(224)
    histRange = (-np.pi, np.pi)
    nhead, edges = np.histogram(gammaFull, density=True, range=histRange, bins=36)
    gammaFullPlt.plot(edges[:-1]+np.diff(edges)/2, nhead, color='grey')
    gammaFullPlt.set_xlim(histRange)
    gammaFullPlt.set_xlabel('rel. heading (full)')
    myAxisTheme(gammaFullPlt)

    headingfig.suptitle(titleString, fontsize=13)
    headingfig.tight_layout()

    try:
        mkdir(datadir + 'analysis'+sep+'heading'+sep)
    except OSError:
        print('Plot directory already exists.')

    headingfig.savefig(datadir+'analysis'+sep+'heading'+sep+FODataFile[0:-4]+'_headingDistribution.pdf', format='pdf')

    # # Stripe tracking trial....

    # # Walking in 2D plane trial.....

    # ## Plot trace and mark object locations
    # (1) plot raw trace .........
    if sceneType == 'plane':
        tStart = 0
        tEnd = len(FOData[:, 1])
        tStep = 72
        frameRange = range(tStart, tEnd, tStep)
        colMap = 'Accent'
        arrowLength = 5

        trajfig = plt.figure(figsize=(10, 10))
        gs = gridspec.GridSpec(2, 1, height_ratios=np.hstack((10, 1)))

        axTraj = trajfig.add_subplot(gs[0]) #trace plot
        axTime = trajfig.add_subplot(gs[1]) #time line

        plotPosInRange(axTraj, axTime, frameRange, FOData[:, 0], FOData[:, 1], FOData[:, 2], np.pi/180*FOData[:, 5],
                       colMap, arrowLength, 0.5, 5)
        axTraj.scatter(visObjCoords[:, 0], visObjCoords[:, 1], 50, alpha=0.75, facecolor='black', edgecolors='none')
        axTraj.scatter(invisObjCoords[:, 0], invisObjCoords[:, 1], 50, alpha=0.5, facecolors='none', edgecolors='black')
        axTraj.set_xlabel(header[1], fontsize=12)
        axTraj.set_ylabel(header[2], fontsize=12)
        axTraj.set_title('Walking trace of ' + titleString)
        axTraj.set_xlim([min(FOData[:, 1]) - 20, max(FOData[:, 1]) + 20])
        axTraj.set_ylim([min(FOData[:, 2]) - 20, max(FOData[:, 2]) + 20])
        myAxisTheme(axTraj)

        axTime.set_xlabel(header[0], fontsize=12)
        plt.xlim((0, FOData[-1, 0]))
        timeAxisTheme(axTime)

        try:
            mkdir(datadir + 'analysis'+sep+'tracePlot'+sep)
        except OSError:
            print('Plot directory already exists.')

        trajfig.savefig(datadir+'analysis'+sep+'tracePlot'+sep+FODataFile[0:-4]+'_traceObjectPlot.pdf', format='pdf')

    # (2) plot collapsed trace ........
    if sceneType == 'plane':
        tStart = 0
        tEnd = numFrames_ds
        tStep = 4
        frameRange = range(tStart, tEnd, tStep)
        colMap = 'Accent'
        arrowLength = 5

        colTrajFig = plt.figure(figsize=(9, 10))
        gs = gridspec.GridSpec(2, 1, height_ratios=np.hstack((10, 1)))

        colTrajFig.suptitle('Collapsed walking trace ("mini arena" with central object)\n' + titleString, fontsize=13)

        axTraj = colTrajFig.add_subplot(gs[0]) # trace plot
        axTime = colTrajFig.add_subplot(gs[1]) # time line
        plotPosInRange(axTraj, axTime, frameRange, time_ds, xPosMA_ds, yPosMA_ds, angle_ds, colMap, 4, 0.5, 7)
        axTraj.plot(0, 0, marker='o', markersize=20, linestyle='none', alpha=0.5, color='black')
        axTraj.set_xlabel(header[1], fontsize=12)
        axTraj.set_ylabel(header[2], fontsize=12)
        axTraj.set_ylim([-arenaRad-5, arenaRad + 5])
        axTraj.set_xlim([-arenaRad-5, arenaRad + 5])
        myAxisTheme(axTraj)
        axTime.set_xlabel(header[0], fontsize=12)
        plt.xlim((0, time_ds[-1]))
        timeAxisTheme(axTime)

        try:
            mkdir(datadir+'analysis'+sep+'collapsedTracePlot'+sep)
        except OSError:
            print('Plot directory already exists.')

        colTrajFig.savefig(datadir+'analysis'+sep+'collapsedTracePlot'+sep+FODataFile[0:-4]+'_traceObjectPlot_ds.pdf',
                           format='pdf')

    # # Save values for future analysis
    if sceneType == 'stripe':
        xPosMA_ds = np.nan*np.ones(np.size(time_ds))
        yPosMA_ds = np.nan*np.ones(np.size(time_ds))

    toSave = {'time': time_ds,
              'xPos': xPos_ds,
              'yPos': yPos_ds,
              'xPosInMiniarena': xPosMA_ds,
              'yPosInMiniarena': yPosMA_ds,
              'headingAngle': angle_ds,
              'rotVelo': vRot_ds,
              'transVelo': vTrans_ds,
              'gammaFull': gammaFull,
              'gamma': gamma,
              'moving': moving}

    np.save(expDir + FODataFile[:-4], toSave)

    plt.close('all')

    return 0