Snakefile.fast

import gzip
import os
import re
import glob
import importlib.util
import scanpy as sc
import anndata


abs_path=os.path.abspath('')

# [A] SCRIPTS IMPORT
spec_u = importlib.util.spec_from_file_location("utility_functions",f"{abs_path}/scripts/utility_functions.py")
utilities = importlib.util.module_from_spec(spec_u)
spec_u.loader.exec_module(utilities)

spec_l = importlib.util.spec_from_file_location("tn_layers",f"{abs_path}/scripts/tn_layers.py")
layers = importlib.util.module_from_spec(spec_l)
spec_l.loader.exec_module(layers)

spec_p = importlib.util.spec_from_file_location("paral_samples",f"{abs_path}/scripts/paral_samples.py")
paral = importlib.util.module_from_spec(spec_p)
spec_p.loader.exec_module(paral)



# [B] CONFIG HANDLING
configfile: f'{abs_path}/config.yaml'

if config['atac']==True:
    exp='atac'
else:
    exp='scget'

include: f'{abs_path}/modules/init_lane.smk'
include: f'{abs_path}/modules/shared_rules.smk'
include: f'{abs_path}/modules/scget_wf.smk.fast'
#include: f'{abs_path}/modules/{exp}_wf.smk'

# [b.1] DEFINITION OF TRANSPOSASES AND BARCODES FOR SCGET ANALYSIS
if config['tn5']!=True:
    TN_BARCODE=config['barcodes']
    TN_BARCODES={'tnh':TN_BARCODE['tnh']}
    BARCODES=TN_BARCODES['tnh']
elif config['tnh']!=True:
    TN_BARCODE=config['barcodes']
    TN_BARCODES={'tn5':TN_BARCODE['tn5']}
    BARCODES=TN_BARCODES['tn5']
else:
    TN_BARCODES=config['barcodes']
    BARCODES=TN_BARCODES['tn5']+TN_BARCODES['tnh']
    
# [b.2] OTHER VARIABLES
CELL_NUMBER=config['cell_number']  
BC_PATTERN=config['bc_pattern']  

INPUT_LIST=config['input_list']
samples_tot=[x.strip() for x in open(f'{INPUT_LIST}').read().split('\n')[:-1]]
SAMPLES=list(set([x.split(' ')[-1] for x in samples_tot if len(x)>0]))

READS= config['reads']
THREADS=config['threads']
INPUT_PATH=config['input_path']
INPUT_LIST=config['input_list']
OUTPUT_PATH=config['output_path']
# [b.3] SELECTION OF OUTPUT MATRICES CHARACTERISTICS
binary_dictionary={True:'-B',False:''}
BINARY=binary_dictionary[config['binary']]

#  create log dir
log_path = OUTPUT_PATH+"/logs_slurm"
try:
    os.mkdir(OUTPUT_PATH+"/logs_slurm")
    os.mkdir(log_path)
except OSError:
    try:
        os.makedirs(log_path)
        print("Successfully created the directory %s " % log_path)
    except OSError:
        print ("Creation of the directory %s failed" % log_path)
else:
    print ("Successfully created the directory %s " % log_path)


# [C] WORKFLOW
rule all:
    input:
        expand('{output}/{sample}/{sample}_stats.txt', output=OUTPUT_PATH,sample=SAMPLES),
        expand('{output}/{sample}/{sample}.h5ad', output=OUTPUT_PATH,sample=SAMPLES)

#rule metrics:
#    input: