#!/usr/bin/env python import time from DurandDB import blastq from matplotlib import pyplot def plot_histogram(br_id=None, nc_id=None, stype=None, color='blue', xlimit=None, ylimit=None, fprefix=None, title=None, ylog=False, set_id_filter=None, symmetric=False): histarr = bq.fetch_score_histogram(br_id=br_id, nc_id=nc_id, stype=stype, set_id_filter=set_id_filter, symmetric = symmetric) pyplot.ioff() # turn off figure updates pyplot.rc('figure', figsize=(8,5)) fig = pyplot.figure() ax = fig.add_subplot(111) #if xlog is not None: # xlog = (stype == 'e_value') if stype == 'e_value': ax.set_xscale('log') if ylog: ax.set_yscale('log') fig.suptitle("%s\nScore Histogram: (%s, %s, %s)" % (title, stype, br_id, nc_id)) # left positions left = [ a[0] for a in histarr ] width = [ a[1] - a[0] for a in histarr ] height = [ a[2] for a in histarr ] pyplot.bar( left, height, width, color=color, edgecolor=color) #pyplot.axis('tight') #if xlimit is not None: ax.set_xlim( *xlimit) #if ylimit is not None: ax.set_ylim( *ylimit) #else: # pyplot.ylim( (min(height),max(height)) ) # for whatever reason, bar() doesn't correctly handle the log scale range #if ylog and ylimit is not None: # pyplot.ylim((0.5, max(height))) if stype == 'e_value': ax.invert_xaxis() ax.grid(color='gray', alpha=0.5) if stype=='e_value': xlabel='-log(e_value / dbsize + 1E-200)' else: xlabel=stype ax.set_xlabel(xlabel) ax.set_ylabel('Count') fig.subplots_adjust(bottom=0.1, left=0.1, top=0.90, right=0.98) if fprefix is not None: fname = fprefix + '_%s_%s_%s_%s.png' % (stype, br_id, nc_id, set_id_filter) fig.savefig( 'figures/%s' % fname) pyplot.close() else: fig.show() return (left, height, width) if __name__ == "__main__": bq = blastq.blastq(cacheq=True) date = time.strftime('%Y%m%d') #plot_histogram( br_id=75, nc_id=698, stype='nc_score', color='blue', # fprefix=date, title='Human_Mouse', xlimit=(-0.01,1.01)) #plot_histogram( br_id=75, nc_id=698, stype='e_value', color='green', # fprefix=date, title='Human_Mouse', ylog=True) #plot_histogram( br_id=87, nc_id=731, stype='nc_score', color='blue', # fprefix=date, title='S_cer', xlimit=(-0.01,1.01)) #plot_histogram( br_id=87, nc_id=731, stype='e_value', color='green', # fprefix=date, title='S_cer', ylog=True) #plot_histogram( br_id=88, nc_id=732, stype='nc_score', color='blue', # fprefix=date, title='Pre-post', xlimit=(-0.01,1.02)) #plot_histogram( br_id=88, nc_id=732, stype='e_value', color='green', # fprefix=date, title='Pre-post', ylog = True) #plot_histogram( br_id=89, nc_id=733, stype='nc_score', color='blue', # fprefix=date, title='Post-post', xlimit=(-0.01,1.05)) #plot_histogram( br_id=89, nc_id=733, stype='e_value', color='green', # fprefix=date, title='Post-post', ylog=True) #plot_histogram( br_id=90, nc_id=734, stype='nc_score', color='blue', # fprefix=date, title='Four_Species', xlimit=(-0.01,1.05)) #plot_histogram( br_id=90, nc_id=734, stype='e_value', color='green', # fprefix=date, title='Four_Species', ylog=True) #plot_histogram( br_id=91, nc_id=735, stype='nc_score', color='blue', # fprefix=date, title='Nine_Species', xlimit=(-0.01,1.05)) #plot_histogram( br_id=91, nc_id=735, stype='e_value', color='green', # fprefix=date, title='Nine_Species', ylog=True) #for (set_id_filter, title) in ( (None, 'All'), (97, 'Human'), (98, 'Mouse'), # (99, 'Scer'), (100, 'Dmel'), (101, 'Arabidopsis')): # plot_histogram( br_id=95, nc_id=736, stype='nc_score', color='blue', # fprefix=date, title='PPOD: ' + title, xlimit=(-0.01,1.05), # set_id_filter = set_id_filter) # plot_histogram( br_id=95, nc_id=736, stype='e_value', color='green', # fprefix=date, title='PPOD: ' + title, ylog=True, # set_id_filter = set_id_filter) #plot_histogram( nc_id=746, stype='nc_score', color='blue', # fprefix=date, title='Cluster->all', ylog=True) #plot_histogram( nc_id=750, stype='nc_score', color='blue', # fprefix=date, title='Cluster->Cluster', ylog=True) #plot_histogram( nc_id=777, stype='nc_score', color='blue', # fprefix=date, title='NC. Non symmetric, compositional=2, 12 species (nc_id 777)', ylog=False) #plot_histogram( nc_id=779, stype='nc_score', color='blue', # fprefix=date, title='NC. Symmetric, compositional=2, 12 species (nc_id 779)', ylog=False) #plot_histogram( nc_id=780, stype='nc_score', color='blue', # fprefix=date, title='NC. Non symmetric, compositional=f, 12 species (nc_id 780)', ylog=False) #plot_histogram( nc_id=781, stype='nc_score', color='blue', # fprefix=date, title='NC. Symmetric, 12 species (nc_id 781)', ylog=True) #plot_histogram( br_id=105, stype='e_value', color='green', # fprefix=date, title='BLAST. Symmetric, 12 species (br_id 105)', # symmetric = True, ylog=True) #plot_histogram( nc_id=808, stype='nc_score', color='blue', # fprefix=date, title='NC. Symmetric, 48 species (nc_id 808)', ylog=True) #plot_histogram( nc_id=816, stype='nc_score', color='blue', # fprefix=date, title='NC. Symmetric, 48 species (nc_id 816). bit_score>=100', ylog=True) #plot_histogram( nc_id=817, stype='nc_score', color='blue', # fprefix=date, title='NC. Symmetric, 48 species (nc_id 817). bit_score>=60', ylog=True) #plot_histogram( br_id=102, stype='bit_score', color='green', # fprefix=date, title='BLAST. Symmetric, 48 species (br_id 102)', # symmetric = True, ylog=True) #plot_histogram( br_id=102, stype='e_value', color='green', # fprefix=date, title='BLAST. Not Symmetric, 48 species (br_id 102)', # symmetric = False, ylog=True) #plot_histogram( br_id=90, stype='bit_score', color='green', # set_id_filter=91, # fprefix=date, title='BLAST. Not Symmetric, (br_id 90, set 91)', # symmetric = False, # ylog=True) #plot_histogram( nc_id=731, stype='nc_score', color='blue', # set_id_filter=91, # fprefix=date, title='nc_id 731, br_id 89, set 91)', # symmetric = False, # ylog=False) plot_histogram( br_id=102, stype='bit_score', color='green', fprefix=date, title='BLAST. Symmetric, 48 genomes. (br_id 102)', symmetric = True, ylog=True) plot_histogram( br_id=102, stype='bit_score', color='green', set_id_filter=109, fprefix=date, title='BLAST. Symmetric, Human and Mouse (br_id 102, set 109)', symmetric = True, ylog=True) plot_histogram( nc_id=808, stype='nc_score', color='blue', fprefix=date, title='NC. 48 genomes. (nc_id 808)', ylog=False) plot_histogram( nc_id=808, stype='nc_score', color='blue', set_id_filter=109, fprefix=date, title='NC. Human and Mouse (nc_id 808, set 109)', ylog=False)