from trnlib.omeORM import *
from pandas import *

from numpy import array
from sqlalchemy import func

from IPython.core.display import HTML

ome = Session()

most_peaks = read_table('arcA_ChIP_seq_only.txt', index_col='gene')

HTML(most_peaks.sort().to_html())

ome.query(ChipPeakGene).filter_by(gene_name='caiT').all()

kiley_discrepancies = {
                       'pdhR':'very weak, insignificant peak',\
                       'gcd': 'very weak, insignificant peak',\
                       'intF_intergenic': 'variable, phage related peak',\
                       'intF': 'variable, phage related peak',\
                       'yaiZ/yaiY': 'weak peak, fell just below threshold',\
                       'lon': 'variable, potential sraA small RNA contamination',\
                       'ybcV_intragenic': 'weak, variable, phage proximal',\
                       'ybdJ': 'very weak, essentially no signal',\
                       'ybaY/tesB': 'three significant out of 12',\
                       'hyaA': 'two significant nitrate peaks',\
                       'csgD': 'two significant anaerobic peaks',\
                       'csgB': 'two significant anaerobic peaks',\
                       'ymgJ/ymjI': 'no signal',\
                       'ycgV': 'two signficant anaerobic peaks',\
                       'ompW': 'two significant anaerobic peaks',\
                       'fnrS': 'no signal',\
                       'CO343': 'phage',\
                       'ybdN': 'two significant anaerobic peaks',\
                       'ydcJ/ydcI': 'very weak, insignificant peaks',\
                       'intergenic': 'all weak and insignificant, 2563400 is phage',\
                       'ryeA/pphA': 'three weak insignificant peaks',\
                       'yegE/udk': 'no signal',\
                       'yohO': 'weak, insignificant signal, possibly small RNA contamination',\
                       'ung/yfiD': 'no signal',\
                       'yhiS_2_intragenic': 'two significant anaerobic peaks',\
                       'gadE': 'two weak peaks, one significant',\
                       'dctA': 'two significant anaerobic peaks',\
                       'mdtL': 'no signal',\
                       'yifK': 'no upstream signal',\
                       'spfl': 'weak signal, possibly small RNA contamination',\
                       'secE': 'ribosomal contamination',\
                       'proP': 'weak, insignificant nitrate peaks',\
                       'cadB': 'two significant anaerobic peaks'  
                       }


len(kiley_discrepancies.keys())