@author: hhuang2
"""
# import glob
import sqlite3 as sql
# from utils import phase_block_check as ps
from utils import IMGTdbIO, CompareSeq
import os
import re
locus = 'DQB1'
#pkl_fp = '../Output/SG39_DRpairs/SG39_HLA_'+ locus +'_paired.pkl'
pkl_fp = '../Output/SG39/2018/SG39_DRpairs/SG39_HLA_' + locus + '_paired.pkl'
DRpair_seqInfo = IMGTdbIO.load_pickle2dict(pkl_fp)
case_count = len(DRpair_seqInfo)
print('Locus ' + locus + ' has ' + str(case_count) + ' paired cases.')
DB_fp = "../Output/SG39/2018/SG39_DRpairs/SG39_HLA_" + locus + "_paired.db"
conn = sql.connect(DB_fp)
cursor = conn.cursor()
cursor.execute('''CREATE TABLE IF NOT EXISTS DR_pair_comparison
(BMT_caseID text, QC text,
PS1_HLATyping text,
PS1_GLstringM text, PS1_SeqM text,
PS2_HLATyping text,
PS2_GLstringM text, PS2_SeqM text,
Audit text, Active text, Comment text)''')
"""
Created on Tue Oct 3 14:42:06 2017
@author: hhuang2
"""
import glob
import sqlite3 as sql
# from utils import phase_block_check as ps
from utils import IMGTdbIO, CompareSeq
import os
import re
fname = '../Output/SG41_52/2018/IMGTv3310/SG41_52_DRpair_Stats/SG41_52_pairedCases_Stats.pkl'
Matching_cases_stats = IMGTdbIO.load_pickle2dict(fname)
## 'All_paired'
groupType = 'fiveLoci_paired' # groupType = 'ClassI_paired' # groupType = 'All_paired'
group_caseIDs = Matching_cases_stats[groupType]
All_loci = ['A', 'B', 'C', 'DRB1', 'DQB1']#, 'DPB1']
ClassI_loci = ['A', 'B', 'C']
ClassII_loci = ['DRB1', 'DQB1']
CaseStats = {}
LocusStats = {}
#MatchStats = {}
for caseID in group_caseIDs:
#
for locus in ClassI_loci:
ARSregion = ['Exon2', 'Exon3']
import glob
import csv
from utils import IMGTdbIO, CompareSeq
groupType = 'fiveLoci_paired' # groupType = 'ClassI_paired' # groupType = 'All_paired' ; 'fiveLoci_paired'
All_loci = ['A', 'B', 'C', 'DRB1', 'DQB1', 'DPB1']
Five_loci = ['A', 'B', 'C', 'DRB1', 'DQB1']
ClassI_loci = ['A', 'B', 'C']
ClassII_loci = ['DRB1', 'DQB1', 'DPB1']
Group_fname = '../Output/SG41_52/2018/IMGTv3310/SG41_52_DRpair_Stats/fiveLoci_paired_Stats_0125_' + groupType + '.pkl'
Stats_Dict = IMGTdbIO.load_pickle2dict(Group_fname)
CaseStats = Stats_Dict['CaseStats']
LocusStats = Stats_Dict['LocusStats']
db_fp= '../Database/'
#key = '83687'
#CaseStats[key]
#key in group_caseIDs
## : paired cases HLA typing stats
fname = '../Output/SG41_52/2018/IMGTv3310/SG41_52_DRpair_Stats/SG41_52_pairedCases_Stats.pkl'
Matching_cases_stats = IMGTdbIO.load_pickle2dict(fname)
group_caseIDs = Matching_cases_stats[groupType]
CaseMatchTable = {}
"""
import glob
import csv
from utils import IMGTdbIO, CompareSeq
groupType = 'fiveLoci_paired' # groupType = 'ClassI_paired' # groupType = 'All_paired' ; 'fiveLoci_paired'
All_loci = ['A', 'B', 'C', 'DRB1', 'DQB1', 'DPB1']
Five_loci = ['A', 'B', 'C', 'DRB1', 'DQB1']
ClassI_loci = ['A', 'B', 'C']
ClassII_loci = ['DRB1', 'DQB1', 'DPB1']
Group_fname = '../Output/SG39/2018/SG39_Stats/fiveLoci_paired_Stats_0125_' + groupType + '.pkl'
Stats_Dict = IMGTdbIO.load_pickle2dict(Group_fname)
CaseStats = Stats_Dict['CaseStats']
LocusStats = Stats_Dict['LocusStats']
db_fp = '../Database/'
#key = '84571'
#CaseStats[key]
#key in group_caseIDs
## : paired cases HLA typing stats
fname = '../Output/SG39/2018/SG39_DRpairs/SG39_pairedCases_Stats.pkl'
Matching_cases_stats = IMGTdbIO.load_pickle2dict(fname)
CaseMatchTable = {}
for locus in All_loci:
@author: hhuang2
"""
from utils import IMGTdbIO #, CompareSeq
from collections import Counter
groupType = 'All_paired' # groupType = 'ClassI_paired' # groupType = 'All_paired'
All_loci = ['A', 'B', 'C', 'DRB1', 'DQB1', 'DPB1']
Five_loci = ['A', 'B', 'C', 'DRB1', 'DQB1']
ClassI_loci = ['A', 'B', 'C']
ClassII_loci = ['DRB1', 'DQB1', 'DPB1']
Group_fname = '../Output/Stats/ClassI_Stats_1003_' + groupType + '.pkl'
Stats_Dict = IMGTdbIO.load_pickle2dict(Group_fname)
CaseStats = Stats_Dict['CaseStats']
LocusStats = Stats_Dict['LocusStats']
## TODO1: paired cases HLA typing stats
fname = '../Output/SG39_DRpairs/SG39_pairedCases_Stats.pkl'
Matching_cases_stats = IMGTdbIO.load_pickle2dict(fname)
AlleleStats = {}
for locus in All_loci:
AlleleStats[locus] = {}
DRpaired_file = '../Output/SG39_DRpairs/SG39_HLA_' + locus + '_paired.pkl'
DRpaired_table = IMGTdbIO.load_pickle2dict(DRpaired_file)
num_total = len(DRpaired_table)