Example of use GEOParse

geoParseExample.py

import GEOparse
import pandas as pd
import matplotlib.pyplot as plt
import numpy
from matplotlib.backends.backend_pdf import PdfPages

#gseNums = ['GSE14860']
#gseNums = ['GSE49278', 'GSE19750', 'GSE10846']
gseNums = ['GSE19417', 'GSE49278', 'GSE19750', 'GSE10846', 'GSE39582', 'GSE35158', 'GSE19422', 'GSE19987', 'GSE32894', 'GSE35158', 'GSE27155', 'GSE33630', 'GSE56303', 'GSE29695', 'GSE10141', 'GSE9843', 'GSE25097', 'GSE32225', 'GSE26566', 'GSE65858', 'GSE39366', 'GSE22138', 'GSE46517', 'GSE71729', 'GSE8607', 'GSE4573', 'GSE29354', 'GSE71118', 'GSE19949', 'GSE15641', 'GSE26253', 'GSE15460', 'GSE21034', 'GSE9891', 'GSE32062', 'GSE72094', 'GSE26939', 'GSE29174', 'GSE31448', 'GSE45725', 'GSE40435', 'GSE14860']

with PdfPages('GSE_boxplots.pdf') as pdf:
    for gse1 in gseNums:
        gse = GEOparse.get_GEO(geo=gse1, destdir="./downloads")

        # Get expression data and metadata matrices
        exprs = []
        gsmNames = []
        metadata = {}
        for gsm_name, gsm in gse.gsms.items():
            if gsm.metadata['type'][0]=='RNA':
                # Expression data
                if len(gsm.table)>0:
                    tmp = gsm.table['VALUE']
                    tmp.index = gsm.table['ID_REF']
                    gsmNames.append(gsm_name)
                    if len(exprs)==0:
                        exprs = tmp.to_frame()
                    else:
                        exprs = pd.concat([exprs,tmp.to_frame()],axis=1)
                
                # Metadata
                for key,value in gsm.metadata.items():
                    #print(key)
                    #print(value)
                    if (key=='characteristics_ch1' or key=='characteristics_ch2') and (len([i for i in value if i!=''])>1 or value[0].find(': ')!=-1):
                        #print(value)
                        tmpVal = 0
                        for tmp in value:
                            splitUp = [i.strip() for i in tmp.split(':')]
                            #print(splitUp)
                            if len(splitUp)==2:
                                if not splitUp[0] in metadata:
                                    metadata[splitUp[0]] = {}
                                metadata[splitUp[0]][gsm_name] = splitUp[1]
                            else:
                                if not key in metadata:
                                    metadata[key] = {}
                                metadata[key][gsm_name] = splitUp[0]
                    else:
                        if not key in metadata:
                            metadata[key] = {}
                        if len(value)==1:
                            metadata[key][gsm_name] = ' '.join([j.replace(',',' ') for j in value])
    
        # Write expression data matrix to file
        exprs.columns = gsmNames
        with open('exprs/'+gse1+'_exprs.csv','w') as outFile:
            exprs.to_csv(outFile)

        # Write metadata matrix to file
        with open('pData/'+gse1+'_pData.csv','w') as outFile:
            outFile.write('Metadata,'+','.join(gsmNames))
            for key in metadata:
                tmp = [key]
                for gsm_name in gsmNames:
                    if gsm_name in metadata[key]:
                        tmp.append(metadata[key][gsm_name])
                    else:
                        tmp.append('NA')
                outFile.write('\n'+','.join(tmp))

        # Plot boxplot of expression data
        plt.boxplot(exprs.transpose(),showfliers=False)
        plt.title(gse1)
        pdf.savefig()
        plt.close()
        try:
            plt.boxplot(numpy.log2(exprs).transpose(),showfliers=False)
            plt.title('log2('+gse1+')')
            pdf.savefig()
            plt.close()
        except:
            pass

        # Write out platform information
        for gpl_name, gpl in gse.gpls.items():
            with open('annot/'+gse1+'_'+gpl_name+'_gpl.csv','w') as outFile:
                gpl.table.to_csv(outFile)