Beispiel #1
0
def fieldtobact(expdat,field,bactname='',meanreads=1000,cutoff=0):
	"""
	convert values in a map file field to a new bacteria (to facilitate numeric analysis)
	input:
	expdat : Experiment
	field : string
		name of the field to convert
	bactname : string
		name of the new bacteria (empty to have similar to field name)
	meanreads : int
		the mean number of reads for the new field bacteria
	cutoff : int
		the minimal value of the field per sample (otherwise replace with meanreads)

	output:
	newexp : Experiment
		with added bacteria with the field vals as reads
	"""
	params=locals()

	if len(bactname)==0:
		bactname=field
	fv=hs.getfieldvals(expdat,field)
	vals=np.array(hs.tofloat(fv))
	okpos=np.where(vals>=cutoff)[0]
	badpos=np.where(vals<cutoff)[0]
	scalefactor=np.mean(vals[okpos])
	vals[okpos]=(vals[okpos]/scalefactor)*meanreads
	vals[badpos]=meanreads
	newexp=hs.copyexp(expdat)
	hs.insertbacteria(newexp,vals,bactname,bactname,logit=False)
	newexp.filters.append('add bacteria from map field %s' % field)
	hs.addcommand(newexp,"fieldtobact",params=params,replaceparams={'expdat':expdat})
	return(newexp)
Beispiel #2
0
def sortsamples(exp,field,numeric=False,logit=True):
	"""
	sort samples according to field
	input:
	exp : Experiment
	field : string
		name of the field to sort by
	numeric : bool
		True for numeric values in field, false for text
	output:
	newexp : Experiment
		the sorted experiment
	"""
	params=locals()

	fvals=hs.getfieldvals(exp,field)
	if numeric:
		fvals=hs.tofloat(fvals)
	svals,sidx=hs.isort(fvals)
	newexp=hs.reordersamples(exp,sidx)

	if logit:
		hs.addcommand(newexp,"sortsamples",params=params,replaceparams={'exp':exp})
		newexp.filters.append('sorted samples by field %s' % field)
	return newexp
Beispiel #3
0
def loaddistmat(expdat,dmfilename):
	"""
	load a distance matrix (from qiime) for analysis
	input:
	expdat : Experiment
	dmfilename : string
		name of the qiime distance matrix file

	output:
	distmat : numpy 2d array
		the distance matrix
	dsamp : dict
		the mapping to position in the experiment for each distmat entry
	"""

	fl=open(dmfilename,'rU')
	# get the column ids
	head=fl.readline().strip('\n')
	ids=head.split('\t')
	ids=ids[1:]
	dist=np.array([])
	snames={}
	for idx,cline in enumerate(fl):
		cline=cline.strip("\n")
		vals=cline.split('\t')
		dist=np.vstack((dist,hs.tofloat(vals[1:]))) if dist.size else np.array(hs.tofloat(vals[1:]))
		if not vals[0]==ids[idx]:
			hs.Debug(9,"strange! line %d row head %s but col head %s" % (idx,vals[0],ids[idx]))
		snames[vals[0]]=idx
	fl.close()
	expkeep=[]
	distorder=[]
	dsamp={}
	for idx,csamp in enumerate(expdat.samples):
		if csamp in snames:
			distorder.append(snames[csamp])
			expkeep.append(idx)
			dsamp[csamp]=snames[csamp]
	hs.Debug(6,"%d samples in dist mat, %d samples in experiment" % (len(ids),len(expdat.samples)))
	hs.Debug(6,"%d samples to keep from dist mat, %d samples to keep from experiment" % (len(distorder),len(expkeep)))
	return dist,dsamp
Beispiel #4
0
def plotseqfreq(expdat,seqs,toaxis=False,xfield=False,normalizey=False):
	"""
	plot the frequency of sequences in seq as a function of the sortfield
	input:
	expdat : Experiment
	seqs : list of sequence strings
		a list of sequnces (acgt) to plot
	toaxis : matplotlib axis
		if not empty - the axis to plot to, False plot a new figure
	xfield : string
		if not empty - space the points on the x axis according to (numeric) value of in xfield, False - just according to the sorted order
	normalizey : bool
		True: normalize all y values to 0-1, False: no normalization
	"""


	if xfield:
		xdat=hs.tofloat(getfieldvals(expdat,xfield))
	else:
		xdat=range(len(expdat.samples))
	sv,si=hs.isort(xdat)
	if not toaxis:
		figure()
		toaxis=plt.gca()
	labels=[]
	for cseq in seqs:
		if cseq not in expdat.seqdict:
			continue
		spos=expdat.seqdict[cseq]
		cdat=expdat.data[spos,si]
		if normalizey:
			cdat=cdat/sum(cdat)
		toaxis.plot(sv,cdat)
		labels.append(str(expdat.sids[spos])+'-'+expdat.tax[spos])
	labels=hs.clipstrings(labels,20,reverse=True)
	toaxis.legend(labels,prop={'size':6})
	toaxis.set_xticks(xdat)
Beispiel #5
0
def plotexp(exp,sortby=False,numeric=False,minreads=4,rangeall=False,seqdb=None,cdb=None,showline=True,ontofig=False,usegui=True,showxall=False,showcolorbar=False,ptitle=False,lowcutoff=1,uselog=True,showxlabel=True,colormap=False,colorrange=False,linewidth=2,subline='',showhline=True,newfig=True,fixfont=False,fontsize=None,nosort=False,zeroisnone=False,xlabelrotation=45,showtaxnames=False):
	"""
	Plot an experiment
	input:
	exp - from load()
	sortby - name of mapping file field to sort by or Flase to not sort
	numeric - True if the field is numeric
	minreads - minimum number of reads per bacteria in order to show it or 0 to show all
	rangeall - True to show all frequencies in image scale, false to saturate at 10%
	seqdb - the SRBactDB database (from bactdb.load)
	cdb - the cool sequences database (from cooldb.load), or None (default) to use the heatsequer loaded cdb
	showline - if True plot lines between category values
	ontofig - name of ontology to plot for bactdb or false to no plot
	usegui - True use a gui for otu summary, False just print
	showxall - True to show all sample names when not sorting, False to show no more than 10
	showcolorbar - True to plot the colorbar. False to not plot
	ptitle : str (optional)
		'' to show o show processing history as name, None to not show title, or str of name of the figure
	lowcutoff - minimal value for read (for 0 log transform) - the minimal resolution - could be 10000*2/origreads
	showxlabel : bool
		True to show the x label (default), False to hide it
	colormap : string or False
		name of colormap or False (default) to use mpl default colormap
	colorrange : [min,max] or False
		[min,max] to set the colormap range, False to use data min,max (default) as specified in rangeall
	subline : str
		Name of category for subline plotting or '' (Default) for no sublines
	showhline : bool
		True (default) to plot the horizontal lines listed in exp.hlines. False to not plot them
	newfig : bool
		True (default) to open figure in new window, False to use current
	fixfont : bool (optional)
		False (default) to use fixedfont, True to use fixed width font
	fontsize : int or None (optional)
		None (default) to use default font size, number to use that font size
	nosort : bool (optional)
		False (default) to sort by the sort field, True to skip the sorting
	zeroisnone : bool (optional)
		False (default) to plot zeros as 0, True to assign None (white color)
	xlabelrotation : int (optional)
		the rotation of the xtick labels
	showtaxnames : book (optional)
		False (default) to not show tax names (need to press 'h' to show)
		True to show the taxonomy names

	output:
	newexp - the plotted experiment (sorted and filtered)
	ax - the plot axis
	"""

	hs.Debug(1,"Plot experiment %s" % exp.studyname)
	hs.Debug(1,"Commands:")
	for ccommand in exp.commands:
		hs.Debug(1,"%s" % ccommand)

	if exp.sparse:
		hs.Debug(9,'Sparse matrix - converting to dense')
		exp=hs.copyexp(exp,todense=True)

	vals=[]
	if cdb is None:
		cdb=hs.cdb
	if seqdb is None:
		seqdb=hs.bdb
	if sortby:
		if not nosort:
			hs.Debug(1,"Sorting by field %s" % sortby)
			for csamp in exp.samples:
				vals.append(exp.smap[csamp][sortby])
			if numeric:
				hs.Debug(1,"(numeric sort)")
				vals=hs.tofloat(vals)
			svals,sidx=hs.isort(vals)
			newexp=hs.reordersamples(exp,sidx)
		else:
			hs.Debug(1,"no sorting but showing columns")
			svals=hs.getfieldvals(exp,sortby)
			newexp=hs.copyexp(exp)
	else:
		hs.Debug(1,"No sample sorting")
		svals=hs.getfieldvals(exp,'#SampleID')
		newexp=hs.copyexp(exp)
	hs.Debug(1,"Filtering min reads. original bacteria - %d" % len(newexp.seqs))
	if minreads>0:
		newexp=hs.filterminreads(newexp,minreads,logit=uselog)
	hs.Debug(1,"New number of bacteria %d" % len(newexp.seqs))
	newexp.seqdb=seqdb
	newexp.cdb=cdb
	newexp.scdb=hs.scdb

	# if usegui:
	# 	hs.Debug(1,"Using the GUI window")
	# 	import heatsequer.plots.plotwingui
	# 	from PyQt4 import QtGui

	# 	app = QtGui.QApplication(sys.argv)
	# 	guiwin = heatsequer.plots.plotwingui.PlotGUIWindow(newexp)

#	ldat=ldat[:,sidx]
	ldat=newexp.data
	if zeroisnone:
		ldat[ldat==0]=None
	if uselog:
		hs.Debug(1,"Using log, cutoff at %f" % lowcutoff)
		ldat[np.where(ldat<lowcutoff)]=lowcutoff
		ldat=np.log2(ldat)
	oldparams=plt.rcParams
	mpl.rc('keymap',back='c, backspace')
	mpl.rc('keymap',forward='v')
	mpl.rc('keymap',all_axes='A')
	if newfig:
		f=plt.figure(tight_layout=True)
	else:
		f=plt.gcf()
	# set the colormap to default if not supplied
	if not colormap:
		colormap=plt.rcParams['image.cmap']
	# plot the image
	if colorrange:
		hs.Debug(1,"colormap range is 0,10")
		iax=plt.imshow(ldat,interpolation='nearest',aspect='auto',clim=colorrange,cmap=plt.get_cmap(colormap))
	elif rangeall:
		hs.Debug(1,"colormap range is all")
		iax=plt.imshow(ldat,interpolation='nearest',aspect='auto',cmap=plt.get_cmap(colormap))
	else:
		hs.Debug(1,"colormap range is 0,10")
		iax=plt.imshow(ldat,interpolation='nearest',aspect='auto',clim=[0,10],cmap=plt.get_cmap(colormap))

	if ptitle is not None:
		if not ptitle:
			hs.Debug(1,"Showing filters in title")
			if (len(newexp.filters))>4:
				cfilters=[newexp.filters[0],'...',newexp.filters[-2],newexp.filters[-1]]
			else:
				cfilters=newexp.filters
			cfilters=hs.clipstrings(cfilters,30)
			ptitle='\n'.join(cfilters)
		plt.title(ptitle,fontsize=10)

	ax=iax.get_axes()
	ax.autoscale(False)

	# plot the sublines (smaller category lines)
	if subline:
		slval=hs.getfieldvals(newexp,subline)
		prevval=slval[0]
		for idx,cval in enumerate(slval):
			if cval!=prevval:
				xpos=idx-0.5
				plt.plot([xpos,xpos],[-0.5,np.size(ldat,0)-0.5],'w:')
				prevval=cval

	if showline:
		hs.Debug(1,"Showing lines")
		labs=[]
		labpos=[]
		linepos=[]
		minpos=0
		svals.append('end')
		for idx,cval in enumerate(svals[:-1]):
			if cval==svals[idx+1]:
				continue
			labpos.append(minpos-0.5+float(idx+1-minpos)/2)
			minpos=idx+1
			linepos.append(idx+0.5)
			labs.append(cval)
		hs.Debug(1,"number of lines is %d" % len(linepos))
		if showxlabel:
			ax.set_xticks(labpos)
			ax.set_xticklabels(labs,rotation=xlabelrotation,ha='right')
		for cx in linepos:
			plt.plot([cx,cx],[-0.5,np.size(ldat,0)-0.5],'k',linewidth=linewidth)
			plt.plot([cx,cx],[-0.5,np.size(ldat,0)-0.5],'w:',linewidth=linewidth)
	else:
		hs.Debug(1,"Not showing lines")
		if showxall or len(newexp.samples)<=10:
			hs.Debug(1,"less than 10 samples, showing all sample names")
			ax.set_xticklabels(svals,rotation=90)
			ax.set_xticks(range(len(newexp.samples)))
	# f.tight_layout()
	ax.set_ylim(-0.5,np.size(ldat,0)-0.5)

	if fixfont:
		fontProperties = {'family':'monospace'}
		ax.set_yticklabels(ax.get_yticks(), fontProperties)

	if showcolorbar:
		hs.Debug(1,"Showing colorbar")
		cb=plt.colorbar(ticks=list(np.log2([2,10,100,500,1000])))
		cb.ax.set_yticklabels(['<0.02%','0.1%','1%','5%','>10%'])

	# create the plot
	ax.expdat=newexp
	ax.lastselect=-1
	ax.sampline=''
	ax.ofig=f
	ax.labelson=False
	ax.labelnames=[]
	f.canvas.mpl_connect('button_press_event', onplotmouseclick)
	f.canvas.mpl_connect('key_press_event', onplotkeyclick)
#	show()
	plt.rcParams=oldparams

	# if want the ontology analysis for a given category:
	if ontofig:
		hs.Debug(1,"Ontofig is set")
		newexp.ontofigname=ontofig
	else:
		newexp.ontofigname=False

	# if we want gui, open it
	if usegui:
		hs.Debug(1,"Using the GUI window")
		import heatsequer.plots.plotwingui
#		from PyQt4 import QtGui

#		app = QtGui.QApplication(sys.argv)
		guiwin = heatsequer.plots.plotwingui.PlotGUIWindow(newexp)
		from heatsequer.plots import plotwingui
		guiwin = plotwingui.PlotGUIWindow(newexp)
		ax.guiwin=guiwin
		guiwin.plotfig=f
		guiwin.plotax=ax
		guiwin.show()
	else:
		ax.guiwin=False
		hs.Debug(7,'Not using gui')

	ax.plot_labelsize=fontsize
	if newexp.plotmetadata:
		hs.Debug(1,"Experiment has metadata attached for plotting (%d points)" % len(newexp.plotmetadata))
		for cmet in newexp.plotmetadata:
			addplotmetadata(newexp,field=cmet[0],value=cmet[1],color=cmet[2],inverse=cmet[3],beforesample=cmet[4])
	if showhline:
		if newexp.hlines:
			for cpos in newexp.hlines:
				plt.plot([0,np.shape(newexp.data)[1]],[cpos-0.5,cpos-0.5],'g')
	plt.show()
	if showtaxnames:
		showtaxonomies(newexp,ax,showdb=False,showcontam=False)

#	if usegui:
#		app.exec_()

	return newexp,ax
Beispiel #6
0
def plotexp(exp,sortby=False,numeric=False,minreads=4,rangeall=False,seqdb=None,cdb=None,showline=True,ontofig=False,usegui=True,showxall=False,showcolorbar=False,ptitle=False,lowcutoff=1,uselog=True,showxlabel=True,colormap=False,colorrange=False):
	"""
	Plot an experiment
	input:
	exp - from load()
	sortby - name of mapping file field to sort by or Flase to not sort
	numeric - True if the field is numeric
	minreads - minimum number of reads per bacteria in order to show it or 0 to show all
	rangeall - True to show all frequencies in image scale, false to saturate at 10%
	seqdb - the SRBactDB database (from bactdb.load)
	cdb - the cool sequences database (from cooldb.load)
	showline - if True plot lines between category values
	ontofig - name of ontology to plot for bactdb or false to no plot
	usegui - True use a gui for otu summary, False just print
	showxall - True to show all sample names when not sorting, False to show no more than 10
	showcolorbar - True to plot the colorbar. False to not plot
	ptitle - name of the figure or False to show processing history as name
	lowcutoff - minimal value for read (for 0 log transform) - the minimal resolution - could be 10000*2/origreads
	showxlabel : bool
		True to show the x label (default), False to hide it
	colormap : string or False
		name of colormap or False (default) to use mpl default colormap
	colorrange : [min,max] or False
		[min,max] to set the colormap range, False to use data min,max (default) as specified in rangeall

	output:
	newexp - the plotted experiment (sorted and filtered)
	ax - the plot axis
	"""

	hs.Debug(1,"Plot experiment %s" % exp.studyname)
	hs.Debug(1,"Commands:")
	for ccommand in exp.commands:
		hs.Debug(1,"%s" % ccommand)
	vals=[]
	if sortby:
		hs.Debug(1,"Sorting by field %s" % sortby)
		for csamp in exp.samples:
			vals.append(exp.smap[csamp][sortby])
		if numeric:
			hs.Debug(1,"(numeric sort)")
			vals=hs.tofloat(vals)
		svals,sidx=hs.isort(vals)
		newexp=hs.reordersamples(exp,sidx)
	else:
		hs.Debug(1,"No sample sorting")
		svals=hs.getfieldvals(exp,'#SampleID')
		newexp=hs.copyexp(exp)
	hs.Debug(1,"Filtering min reads. original bacteria - %d" % len(newexp.seqs))
	if minreads>0:
		newexp=hs.filterminreads(newexp,minreads,logit=uselog)
	hs.Debug(1,"New number of bacteria %d" % len(newexp.seqs))
	newexp.seqdb=seqdb
	newexp.cdb=cdb

#	ldat=ldat[:,sidx]
	ldat=newexp.data
	if uselog:
		hs.Debug(1,"Using log, cutoff at %f" % lowcutoff)
		ldat[np.where(ldat<lowcutoff)]=lowcutoff
		ldat=np.log2(ldat)
	oldparams=plt.rcParams
	mpl.rc('keymap',back='c, backspace')
	mpl.rc('keymap',forward='v')
	mpl.rc('keymap',all_axes='A')
	f=figure()
	# set the colormap to default if not supplied
	if not colormap:
		colormap=plt.rcParams['image.cmap']
	# plot the image
	if colorrange:
		hs.Debug(1,"colormap range is 0,10")
		iax=imshow(ldat,interpolation='nearest',aspect='auto',clim=colorrange,cmap=plt.get_cmap(colormap))
	elif rangeall:
		hs.Debug(1,"colormap range is all")
		iax=imshow(ldat,interpolation='nearest',aspect='auto',cmap=plt.get_cmap(colormap))
	else:
		hs.Debug(1,"colormap range is 0,10")
		iax=imshow(ldat,interpolation='nearest',aspect='auto',clim=[0,10],cmap=plt.get_cmap(colormap))

	if not ptitle:
		hs.Debug(1,"Showing filters in title")
		if (len(newexp.filters))>4:
			cfilters=[newexp.filters[0],'...',newexp.filters[-2],newexp.filters[-1]]
		else:
			cfilters=newexp.filters
		cfilters=hs.clipstrings(cfilters,30)
		ptitle='\n'.join(cfilters)
	title(ptitle,fontsize=10)

	ax=iax.get_axes()
	ax.autoscale(False)
	if showline:
		hs.Debug(1,"Showing lines")
		labs=[]
		labpos=[]
		linepos=[]
		minpos=0
		svals.append('end')
		for idx,cval in enumerate(svals[:-1]):
			if cval==svals[idx+1]:
				continue
			labpos.append(minpos-0.5+float(idx+1-minpos)/2)
			minpos=idx+1
			linepos.append(idx+0.5)
			labs.append(cval)
		hs.Debug(1,"number of lines is %d" % len(linepos))
		if showxlabel:
			ax.set_xticks(labpos)
			ax.set_xticklabels(labs,rotation=45,ha='right')
		for cx in linepos:
			plot([cx,cx],[-0.5,np.size(ldat,0)-0.5],'k',linewidth=2)
	else:
		hs.Debug(1,"Not showing lines")
		if showxall or len(newexp.samples)<=10:
			hs.Debug(1,"less than 10 samples, showing all sample names")
			ax.set_xticklabels(svals,rotation=90)
			ax.set_xticks(range(len(newexp.samples)))
	tight_layout()
	ax.set_ylim(-0.5,np.size(ldat,0)+0.5)

	if showcolorbar:
		hs.Debug(1,"Showing colorbar")
		cb=colorbar(ticks=list(np.log2([2,10,100,500,1000])))
		cb.ax.set_yticklabels(['<0.02%','0.1%','1%','5%','>10%'])

	# create the plot
	ax.expdat=newexp
	ax.lastselect=-1
	ax.sampline=''
	ax.ofig=f
	ax.labelson=False
	ax.labelnames=[]
	f.canvas.mpl_connect('button_press_event', onplotmouseclick)
	f.canvas.mpl_connect('key_press_event', onplotkeyclick)
#	show()
	plt.rcParams=oldparams

	# if want the ontology analysis for a given category:
	if ontofig:
		hs.Debug(1,"Ontofig is set")
		newexp.ontofigname=ontofig
	else:
		newexp.ontofigname=False

	# if we want gui, open it
	if usegui:
		hs.Debug(1,"Using the GUI window")
		import heatsequer.plots.plotwingui
		guiwin = heatsequer.plots.plotwingui.PlotGUIWindow(newexp)
#		from heatsequer.plots import plotwingui
#		guiwin = plotwingui.PlotGUIWindow(newexp)
		ax.guiwin=guiwin
		guiwin.plotfig=f
		guiwin.plotax=ax
		guiwin.show()
	else:
		ax.guiwin=False
		hs.Debug(7,'Not using gui')

	if newexp.plotmetadata:
		hs.Debug(1,"Experiment has metadata attached for plotting (%d points)" % len(newexp.plotmetadata))
		for cmet in newexp.plotmetadata:
			addplotmetadata(newexp,field=cmet[0],value=cmet[1],color=cmet[2],inverse=cmet[3],beforesample=cmet[4])
	show()
	return newexp,ax