def __init__(self, rapi_plugin_id, paired=True):
self._plugin = pyrapi.load_aligner(rapi_plugin_id)
self._opts = HiRapiOpts(self._plugin)
self._plugin.init(self._opts._rapi_opts)
self._batch = self._plugin.read_batch(2 if paired else 1)
self._aligner = None
self._ref = None
self._qoffset = self.Qenc_Sanger
def __init__(self, rapi_plugin_id, paired=True, opts=None):
self._plugin = pyrapi.load_aligner(rapi_plugin_id)
self._opts = opts if opts else HiRapiOpts(self._plugin)
self._plugin.init(self._opts._rapi_opts)
self._batch = self._plugin.read_batch(2 if paired else 1)
self._aligner = None
self._ref = None
self._qoffset = self.Qenc_Sanger
def main(argv=None):
start_time = time.time()
batch_size = 100000 # n reads
assert batch_size % 2 == 0
options = parse_args(argv)
plugin = pyrapi.load_aligner('rapi_bwa')
opts = plugin.opts()
#opts.share_ref_mem = True
plugin.init(opts)
opts.n_threads = options.nthreads
_log.info("Using the %s aligner plugin, aligner version %s, plugin version %s",
plugin.aligner_name(), plugin.aligner_version(), plugin.plugin_version())
_log.info("Number of threads: %s", opts.n_threads)
pe = not options.se
_log.info("Working in %s mode", 'paired-end' if pe else 'single-end')
_log.info("Reading data in %s format", options.format)
_log.info("Loading reference %s", options.ref)
ref = plugin.ref(options.ref)
_log.info("Reference loaded")
batch = plugin.read_batch(2 if pe else 1)
# allocate space for reads
batch.reserve(100000 / batch.n_reads_per_frag)
aligner = plugin.aligner(opts)
# print SAM header
print plugin.format_sam_hdr(ref)
input_generator = create_input(options)
def _load_batch():
batch.clear()
_log.info('loading batch %s', batch_count)
for reads in input_generator:
read1 = reads[0]
batch.append(read1['id'], read1['seq'], read1['q'], plugin.QENC_SANGER)
if pe:
read2 = reads[1]
batch.append(read2['id'], read2['seq'], read2['q'], plugin.QENC_SANGER)
if len(batch) >= batch_size:
break
# return whether or not the batch is empty
return len(batch) != 0
def _process_batch():
_log.info("aligning...")
aligner.align_reads(ref, batch)
_log.info("finished aligning. Printing output")
for idx in xrange(batch.n_fragments):
sam = plugin.format_sam_from_batch(batch, idx)
print sam
done = False
batch_count = 0
while not done:
batch_count += 1
_log.info("Batch %s", batch_count)
has_data = _load_batch()
if has_data:
_process_batch()
else:
done = True
ref.unload()
end_time = time.time()
_log.info("Total runtime: %0.3f seconds", end_time - start_time)
