def __init__(self, params, logfile):
"""
Optional parameters:
* parse_output (bool, True): Whether the sieve should write its own output file.
Optional parameters passed to SGA:
* max_edges (int, 400)
* numcpu (int, 4)
* min_branch_length (int, 35)
* min_merge_overlap (int, 15)
* min_assembly_overlap (int, 5)
* max_gap_divergence (int, 0)
* resolve_small (int, 500)
* error_rate (float, 0.02)
"""
param_names = [('parse_output', True), ('max_edges', 400),
('numcpu', 4), ('min_branch_length', 35),
('min_merge_overlap', 15), ('min_assembly_overlap', 5),
('max_gap_divergence', 0), ('resolve_small', 500),
('error_rate', 0.02)]
Sieve.__init__(self,
params,
logfile,
name='SGA',
param_names=param_names)
def __init__(self, params, logfile):
"""
Optional parameters:
* item_limit (int, default 0): Maximum number of sequences to parse. 0 means no limit.
"""
Sieve.__init__(self,
params,
logfile,
name='FASTA translator',
param_names=[('item_limit', 0)])
def __init__(self, params, logfile):
"""
Mandatory parameters:
* sieve (module): Sieve to run
* params (dict): Parameter set listing the different values for each parameter on the following form::
{
'param1': [1, 2],
'param2': ['A', 'B', 'C']
}
The supplied sieve will run once for each possible combination of parameters. The preceding example will thus run 2*3=6 times.
"""
param_names = ['sieve', 'params']
Sieve.__init__(self,
params,
logfile,
name='MultiRunner',
param_names=param_names)
def __init__(self, params, logfile):
"""
Mandatory parameters:
* model_path (str): Path to .hmm model file to use for hmmsearch.
Optional parameters:
* hmmsearch_out(str, 'hmmsearch_out'): Path to temporary hmmsearch output file.
* numcpu (int, 4): Number of threads to use for hmmsearch
* write_only_domain (bool, True): If output FASTA file should contain entire input sequence or just matching domain.
* use_heuristics (bool, True): Heuristics on/off; there is little reason not to use heuristics, HMMer has a higher propensity for crashing if not used and it only increases the number of really low scoring hits anyway,
* longseqdef (int, 151): Sequences longer than this will be considered "long" and are thus compared to `longseqcutoff` instead of `classificationfunction`.
* longseqcutoff (float, 109.64): The classification cutoff (minimum domain score) for long sequences. That is, "long" sequences will only pass if they have a score higher than this.
* minscore (int, 0): Minimum domain score, regardless of length, to pass a sequence.
* min_sequence_length (int, 20): Only consider sequences longer than this.
* max_domain_length (int, 21844): Do not include sequences where the maximum domain is longer than this.
* max_sequence_length (int, 21844000): Do not include sequences longer than this.
* classificationfunction (function, lambda L: self.classifyK*L + self.classifyM): Classification function. The domain score of sequences shorter than `longseqdef` are compared to the result of this function, with the sequence length fed as the parameter. Only those with a domain score higher will pass.
* classifyK (float, 0.7778): Parameter to the default `classificationfunction`.
* classifyM (float, -7.89): Parameter to the default `classificationfunction`.
"""
param_names = [
'model_path',
('hmmsearch_out', 'hmmsearch_out'),
('numcpu', 4),
('write_only_domain', True),
('use_heuristics', True),
('longseqdef', 151),
('longseqcutoff', 109.64),
('minscore', 0),
('min_sequence_length', 20), #minimum fragment length allowed.
('max_domain_length', 21844),
('max_sequence_length', 21844000),
('classifyK', 0.7778),
('classifyM', -7.89),
('classificationfunction', lambda L: self.classifyK*L + self.classifyM)
]
Sieve.__init__(self, params, logfile, name='HMMer hmmsearch', param_names=param_names)
