"*.csfasta.F3.gz",
)
SEQUENCEFILES = tuple([os.path.join(DATADIR, suffix_name)
for suffix_name in SEQUENCESUFFIXES])
SEQUENCEFILES_REGEX = regex(
r"(\S+)-(\S+)-(\S+).(?P<suffix>fastq.1.gz|fastq.gz|sra)")
Sample = PipelineTracks.AutoSample
Sample.attributes = ('tissue', 'condition', 'replicate')
TRACKS = PipelineTracks.Tracks(Sample).loadFromDirectory(
[y for x in SEQUENCESUFFIXES for y in glob.glob(x)],
"(\S+).(fastq.1.gz|fastq.gz|sra)")
EXPERIMENTS = PipelineTracks.Aggregate(TRACKS, labels=("tissue", "condition"))
CONDITIONS = PipelineTracks.Aggregate(TRACKS, labels=("condition", ))
REPLICATES = PipelineTracks.Aggregate(TRACKS, labels=("replicate", ))
#########################################################################
# summarise read 3'
#########################################################################
@follows(mkdir("sequence_characteristics.dir"))
@transform(SEQUENCEFILES,
SEQUENCEFILES_REGEX,
r"sequence_characteristics.dir/\1-\2-\3.\g<suffix>_start.tsv")
def summariseReadStart(infile, outfile):
# this only works for fastq files. Fails with .sra files
# this function and the next section should be replaced with a call to
"pipeline_annotations.py")
###################################################################
###################################################################
# Helper functions mapping tracks to conditions, etc
###################################################################
import CGATPipelines.PipelineTracks as PipelineTracks
Sample = PipelineTracks.AutoSample
# collect sra nd fastq.gz tracks
TRACKS = PipelineTracks.Tracks(Sample).loadFromDirectory(
glob.glob("*.bam"), "(\S+).bam")
# group by experiment (assume that last field is a replicate identifier)
EXPERIMENTS = PipelineTracks.Aggregate(TRACKS, labels=("condition", "tissue"))
GENESETS = PipelineTracks.Tracks(Sample).loadFromDirectory(
glob.glob("*.gtf.gz"), "(\S+).gtf.gz")
###################################################################
###################################################################
###################################################################
def connect():
'''connect to database.
This method also attaches to helper databases.
'''