def get_max_reads_length(reads_file, log, num_checked): file_type = SeqIO.get_read_file_type(reads_file) if not file_type: error('Incorrect extension of reads file: ' + reads_file, log) max_reads_length = max([len(rec) for rec in itertools.islice(SeqIO.parse(SeqIO.Open(reads_file, "r"), file_type), num_checked)]) log.info(reads_file + ': max reads length: ' + str(max_reads_length)) return max_reads_length
def check_file_not_empty(input_filename, message="", log=None): filename = abspath(expanduser(input_filename)) file_type = get_read_file_type(input_filename, log) if (file_type == 'bam'): return try: reads_iterator = SeqIO.parse(SeqIO.Open(filename, "r"), file_type) if next(reads_iterator, None) is None: error("file is empty: %s (%s)" % (filename, message), log=log) except Exception as inst: error(inst.args[0].format(FILE=filename) + "\n\n" + traceback.format_exc().format(FILE=filename), log=log)
def get_max_reads_length(reads_file, log, num_checked): if reads_file in options_storage.dict_of_prefixes: ext = options_storage.dict_of_prefixes[reads_file] file_type = SeqIO.get_read_file_type(ext) else: file_type = SeqIO.get_read_file_type(reads_file) if not file_type: error("incorrect extension of reads file: %s" % reads_file, log) max_reads_length = max([ len(rec) for rec in itertools.islice( SeqIO.parse(SeqIO.Open(reads_file, "r"), file_type), num_checked) ]) log.info("%s: max reads length: %s" % (reads_file, str(max_reads_length))) return max_reads_length
def get_max_reads_length(reads_file, log, num_checked): file_type = get_read_file_type(reads_file, log) max_reads_length = 0 try: max_reads_length = max([ len(rec) for rec in itertools.islice( SeqIO.parse(SeqIO.Open(reads_file, "r"), file_type), num_checked) ]) except Exception as inst: error(inst.args[0].format(FILE=reads_file) + "\n\n" + traceback.format_exc().format(FILE=reads_file), log=log) else: log.info("%s: max reads length: %s" % (reads_file, str(max_reads_length))) return max_reads_length