comparing LAA reports

gistfile1.txt

Python 2.7.9 (default, May 31 2015, 05:04:28)
[GCC 4.8.4] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> from pbcommand.pb_io.report import load_report_from_json
>>> rpt_old = load_report_from_json("/pbi/dept/secondary/siv/smrtlink/smrtlink-alpha/smrtsuite/userdata/jobs_root/000/000212/tasks/pbreports.tasks.amplicon_analysis_consensus-0/consensus_report.json")
>>> rpt_new = load_report_from_json("/pbi/dept/secondary/siv/smrtlink/smrtlink-alpha/smrtsuite/userdata/jobs_root/000/000215/tasks/pbreports.tasks.amplicon_analysis_consensus-0/consensus_report.json")
>>> table_old = rpt_old.tables[0]
>>> table_new = rpt_new.tables[0]
>>> table_old.columns
[<Column id:barcodename header:Barcode nvalues:370 >, <Column id:coarsecluster header:Sequence Cluster nvalues:370 >, <Column id:phase header:Sequence Phase nvalues:370 >, <Column id:sequencelength header:Length (bp) nvalues:370 >, <Column id:predictedaccuracy header:Estimated Accuracy nvalues:370 >, <Column id:totalcoverage header:Subreads coverage nvalues:370 >]
>>> acc_old = table_old.columns[-2].values
>>> acc_new = table_new.columns[-2].values
>>> array(acc_old).mean()
0.99830650270270282
>>> array(acc_new).mean()
0.99889366219839137
>>> bc_old = table_old.columns[0].values
>>> bc_new = table_new.columns[0].values
>>> set(bc_old).difference(set(bc_new))
set([u'180--180', u'378--378'])
>>> set(bc_new).difference(set(bc_old))
set([u'19--19', u'349--349', u'362--362', u'361--361', u'370--370'])
>>> len(bc_old)
370
>>> len(bc_new)
373