parsing Goldsmith uncertainty data to DFT_QCI_thermo library

Parse Goldsmith Data

This gist parses the tab separated text file 'goldsmith_table3.tsv', adds the uncertainty data to the RMG thermo library 'DFT_QC_thermo.py' and saves it to a new library 'DFT_QC_thermo_new.py'

goldsmith_table3.tsv

parse.py

'''
Created on Aug 20, 2014

@author: nickvandewiele
'''
import csv, os
import numpy as np
from rmgpy.data.thermo import ThermoDatabase, ThermoLibrary

path_to_RMG_thermo_db = '/path/to/RMG-database/input/thermo/'

def adjust_non_zero(number):
    return 0.05 if (number == 0.0) else number
    
def parse_uncertainty_data():
    '''
    Returns:
    array with each element a species of the database
    Element is a dictionary with 3 keys:
        1) 'H'
        2) 'S'
        3) 'Cp'
        
    The 'Cp' key has a value that is an array of 7 uncertainties
    (300, 400, 500, 600, 800, 1000, 1500)K
    The other keys have one uncertainty.
    
    Assumes that units of uncertainty are in cal, mol, K.
    
    '''
    uncertainty_data = []
    with open("goldsmith_parsed.tsv") as tsv:
        for line in csv.reader(tsv, delimiter="\t"):
            uncertainty = {}
            uncertainty['H'] = adjust_non_zero(float(line[1]))
            uncertainty['S'] = adjust_non_zero(float(line[3]))
            temp = [line[5],line[7],line[9],line[11], line[13],line[15],line[17]]
            temp = np.array([adjust_non_zero(float(t)) for t in temp])
            uncertainty['Cp'] = temp

            uncertainty_data.append(uncertainty)

    return uncertainty_data

def add_uncertainty_data(uncertainty_data, new_lib_name):
    '''
    The indeces in the original DFT_QCI_thermo library correspond to the number in the original paper
    (J.Phys.Chem.A 2012 116 9033 9057) 
    until index 33. Species 33 in the original paper is not included (C2, singlet), but the indeces
    in the thermo library continued counting.
    
    Therefore, I updated the indeces of the newly created thermo library by omitting index 33.
    
    '''
    global path_to_RMG_thermo_db
    thermoLibraryName = 'DFT_QCI_thermo'
    
    
    t = ThermoDatabase()
    t.load(path_to_RMG_thermo_db, [thermoLibraryName], depository=False) 
    lib = t.libraries[thermoLibraryName]
    
    index_where_offset_starts = 33
    
    for entry in lib.entries.values():
        index = entry.index
        data = entry.data
        if index < index_where_offset_starts:
            index_of_uncertainty_data = index-1
        else:
            index_of_uncertainty_data = index
            entry.index  = index+1
            
        data.H298.uncertainty = uncertainty_data[index_of_uncertainty_data]['H']
        data.S298.uncertainty = uncertainty_data[index_of_uncertainty_data]['S']
        data.Cpdata.uncertainty = uncertainty_data[index_of_uncertainty_data]['Cp']
        
    updated_lib_path = os.path.join(path_to_RMG_thermo_db, 'libraries', new_lib_name+'.py')
    lib.save(updated_lib_path)
    
    print 'Updated library saved to: {0}'.format(updated_lib_path)

def verify(new_lib_name):
    global path_to_RMG_thermo_db    
    
    t = ThermoDatabase()
    t.load(path_to_RMG_thermo_db, [new_lib_name], depository=False)
    
    assert isinstance(t.libraries[new_lib_name], ThermoLibrary)
    assert len(t.libraries[new_lib_name].entries.keys()) != 0

    
    print 'Checked!'
    
if __name__ == '__main__':
    uncertainty_data = parse_uncertainty_data()
    
    print 'Uncertainty data parsed!'
    
    new_lib_name = 'DFT_QCI_thermo_new'
    add_uncertainty_data(uncertainty_data, new_lib_name)
    
    verify(new_lib_name)
    
    print 'Done!'