comparison metfrag.py @ 4:eb581a101672 draft

planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4

3:5ee936e570a7

parser.add_argument('--pctexplpeak_thrshld', default=0)
parser.add_argument('--schema')
parser.add_argument('--cores_top_level', default=1)
parser.add_argument('--chunks', default=1)
parser.add_argument('--meta_select_col', default='name')

parser.add_argument('--ScoreSuspectLists', default='')
parser.add_argument('--MetFragScoreTypes', default="FragmenterScore,OfflineMetFusionScore")
parser.add_argument('--MetFragScoreWeights', default="1.0,1.0")

args = parser.parse_args()
print(args)

# Create temporary working directory
if args.temp_dir:
    wd = args.temp_dir
else:

    '[M+ACN+H]+': 42.033823,
    '[M+ACN+Na]+': 64.015765,
    '[M+2ACN+H]+': 83.06037,
    '[M-H]-': -1.007276,
    '[M+Cl]-': 34.969402,
}

# function to extract the meta data using the regular expressions
def parse_meta(meta_regex, meta_info={}):
    for k, regexes in six.iteritems(meta_regex):

            outfile.write(p[0] + "\t" + p[1] + "\n")

    # =============== Update param based on MSP metadata ======================
    # Replace param details with details from MSP if required
    if 'precursor_type' in meta_info and meta_info['precursor_type'] in adduct_types:

        nm = float(meta_info['precursor_mz']) + adduct_types[meta_info['precursor_type']]
        paramd["PrecursorIonMode"] = int(round(adduct_types[meta_info['precursor_type']], 0))
    else:

        paramd["PrecursorIonMode"] = paramd['PrecursorIonModeDefault']
        nm = float(meta_info['precursor_mz']) + paramd['nm_mass_diff_default']

    paramd["NeutralPrecursorMass"] = nm

    # =============== Create CLI cmd for metfrag ===============================
    cmd = "metfrag"
    for k, v in six.iteritems(paramd):
        if not k in ['PrecursorIonModeDefault', 'nm_mass_diff_default', 'additional_details']:
            cmd += " {}={}".format(str(k), str(v))

    # =============== Run metfrag ==============================================
    # Filter before process with a minimum number of MS/MS peaks
    if plinesread >= float(args.minMSMSpeaks):

        if int(args.cores_top_level) == 1:
            os.system(cmd)

        elif plinesread and plinesread == pnumlines:
            # =============== Get sample name and additional details for output =======
            spectrac += 1
            paramd, cmd = run_metfrag(meta_info, peaklist, args, wd, spectrac, adduct_types)

            paramds[paramd["SampleName"]] = paramd
            cmds.append(cmd)

            meta_info = {}
            pnumlines = 0
            plinesread = 0

    # end of file. Check if there is a MSP spectra to run metfrag on still
    if plinesread and plinesread == pnumlines:

        paramd, cmd = run_metfrag(meta_info, peaklist, args, wd, spectrac+1, adduct_types)

        paramds[paramd["SampleName"]] = paramd
        cmds.append(cmd)





# Perform multiprocessing on command line call level

4:eb581a101672

parser.add_argument('--pctexplpeak_thrshld', default=0)
parser.add_argument('--schema')
parser.add_argument('--cores_top_level', default=1)
parser.add_argument('--chunks', default=1)
parser.add_argument('--meta_select_col', default='name')
parser.add_argument('--skip_invalid_adducts',  action='store_true')

parser.add_argument('--ScoreSuspectLists', default='')
parser.add_argument('--MetFragScoreTypes', default="FragmenterScore,OfflineMetFusionScore")
parser.add_argument('--MetFragScoreWeights', default="1.0,1.0")


args = parser.parse_args()
print(args)

if os.stat(args.input_pth).st_size == 0:
    exit('Input file empty')


# Create temporary working directory
if args.temp_dir:
    wd = args.temp_dir
else:

    '[M+ACN+H]+': 42.033823,
    '[M+ACN+Na]+': 64.015765,
    '[M+2ACN+H]+': 83.06037,
    '[M-H]-': -1.007276,
    '[M+Cl]-': 34.969402,
    '[M+HCOO]-': 44.99819,
    '[M-H+HCOOH]-': 44.99819,  # same as above but different style of writing adduct
    '[M+CH3COO]-': 59.01385,
    '[M-H+CH3COOH]-': 59.01385  # same as above but different style of writing adduct
}

# function to extract the meta data using the regular expressions
def parse_meta(meta_regex, meta_info={}):
    for k, regexes in six.iteritems(meta_regex):

            outfile.write(p[0] + "\t" + p[1] + "\n")

    # =============== Update param based on MSP metadata ======================
    # Replace param details with details from MSP if required
    if 'precursor_type' in meta_info and meta_info['precursor_type'] in adduct_types:
        nm = float(meta_info['precursor_mz']) - adduct_types[meta_info['precursor_type']]
        paramd["PrecursorIonMode"] = int(round(adduct_types[meta_info['precursor_type']], 0))
    elif not args.skip_invalid_adducts:
        paramd["PrecursorIonMode"] = paramd['PrecursorIonModeDefault']
        nm = float(meta_info['precursor_mz']) - paramd['nm_mass_diff_default']
    else:
        print('Skipping {}'.format(paramd["SampleName"]))
        return '', ''

    paramd["NeutralPrecursorMass"] = nm

    # =============== Create CLI cmd for metfrag ===============================
    cmd = "metfrag"
    for k, v in six.iteritems(paramd):
        if not k in ['PrecursorIonModeDefault', 'nm_mass_diff_default', 'additional_details']:
            cmd += " {}={}".format(str(k), str(v))

    # =============== Run metfrag ==============================================
    print(cmd)
    # Filter before process with a minimum number of MS/MS peaks
    if plinesread >= float(args.minMSMSpeaks):

        if int(args.cores_top_level) == 1:
            os.system(cmd)

        elif plinesread and plinesread == pnumlines:
            # =============== Get sample name and additional details for output =======
            spectrac += 1
            paramd, cmd = run_metfrag(meta_info, peaklist, args, wd, spectrac, adduct_types)

            if paramd:
                paramds[paramd["SampleName"]] = paramd
                cmds.append(cmd)

            meta_info = {}
            pnumlines = 0
            plinesread = 0

    # end of file. Check if there is a MSP spectra to run metfrag on still
    if plinesread and plinesread == pnumlines:

        paramd, cmd = run_metfrag(meta_info, peaklist, args, wd, spectrac+1, adduct_types)

        if paramd:
            paramds[paramd["SampleName"]] = paramd
            cmds.append(cmd)




# Perform multiprocessing on command line call level