From f7fe919ab8252b764868d5feeed4150ec8103931 Mon Sep 17 00:00:00 2001
From: Laurent Modolo <laurent.modolo@ens-lyon.fr>
Date: Wed, 22 May 2019 14:21:45 +0200
Subject: [PATCH] rmi_splitter.py: change ntuple to start from 0
---
src/rmi_splitter/rmi_splitter.py | 32 +++++-----
src/rmi_splitter/tests/data_test.py | 12 ++--
src/rmi_splitter/tests/rmi_splitter_test.py | 69 ++++++++++++++-------
3 files changed, 70 insertions(+), 43 deletions(-)
diff --git a/src/rmi_splitter/rmi_splitter.py b/src/rmi_splitter/rmi_splitter.py
index ad9a77a..095931c 100644
--- a/src/rmi_splitter/rmi_splitter.py
+++ b/src/rmi_splitter/rmi_splitter.py
@@ -133,7 +133,7 @@ def update_config(config):
return config
-def test_adaptator(config, adaptator, ntuple=0, verbose=False):
+def test_adaptator(config, adaptator, ntuple, verbose=False):
"""
Run tests on the adaptator
@@ -237,7 +237,7 @@ def extract_barcode_pos(reads, start, stop, header):
}
-def extract_barcode(reads, config, adaptator, ntuple=0, verbose=False):
+def extract_barcode(reads, config, adaptator, ntuple, verbose=False):
"""
Extract barcode from config from adaptator.
@@ -258,7 +258,7 @@ def extract_barcode(reads, config, adaptator, ntuple=0, verbose=False):
header=coords['header'])
-def write_umi_in_header(reads, config, adaptator, ntuple=0, verbose=False):
+def write_umi_in_header(reads, config, adaptator, ntuple, verbose=False):
"""
Copy the UMI in the header separated by an _ to use later with UMI_tools.
@@ -273,7 +273,7 @@ def write_umi_in_header(reads, config, adaptator, ntuple=0, verbose=False):
return reads
-def list_adaptator_barcode(config, adaptator, ntuple=0, verbose=False):
+def list_adaptator_barcode(config, adaptator, ntuple, verbose=False):
"""
Create a list of concatened barecode seq and a list of concatened barcode
names from the config.
@@ -327,7 +327,7 @@ def create_barcode_dictionaries(config, mismatch=None):
return adaptators_dict
-def match_barcode(reads, config, adaptator, barcode_dictionary, ntuple=0,
+def match_barcode(reads, config, adaptator, barcode_dictionary, ntuple,
verbose=False):
"""
Search barcode suffixtree.
@@ -370,7 +370,7 @@ def match_barcode(reads, config, adaptator, barcode_dictionary, ntuple=0,
str(ntuple))
-def match_barcodes(reads, config, barcode_dictionary, ntuple=0, verbose=False):
+def match_barcodes(reads, config, barcode_dictionary, ntuple, verbose=False):
"""
Search all barcodes
@@ -467,7 +467,7 @@ def remove_barcode_pos(reads, start, stop, header):
return reads
-def remove_barcode(reads, config, adaptator, ntuple=0, verbose=False):
+def remove_barcode(reads, config, adaptator, ntuple, verbose=False):
"""
Remove barcode from sequence (i.e trim read).
@@ -476,7 +476,7 @@ def remove_barcode(reads, config, adaptator, ntuple=0, verbose=False):
params: adaptator
"""
if verbose:
- test_adaptator(config=config, adaptator=adaptator, ntuple=0,
+ test_adaptator(config=config, adaptator=adaptator, ntuple=ntuple,
verbose=verbose)
coords = config[adaptator]['coords']
if adaptator == 'UMI':
@@ -495,7 +495,7 @@ def remove_barcode(reads, config, adaptator, ntuple=0, verbose=False):
)
-def update_position(config, adaptator, adapt, adaptator_length, ntuple=0):
+def update_position(config, adaptator, adapt, adaptator_length, ntuple):
"""
Update barcode position in config file when a barcode is removed
@@ -519,7 +519,7 @@ def update_position(config, adaptator, adapt, adaptator_length, ntuple=0):
return config
-def update_positions(config, adaptator, ntuple=0):
+def update_positions(config, adaptator, ntuple):
"""
Update barcode position in config file when a barcode is removed
@@ -543,7 +543,7 @@ def update_positions(config, adaptator, ntuple=0):
return config
-def remove_barcodes(reads, config, ntuple=0, verbose=False):
+def remove_barcodes(reads, config, ntuple, verbose=False):
"""
Remove barcodes from sequence (i.e trim read).
can be call once by read, otherwise adaptator coords doesn't macht anymore
@@ -559,7 +559,7 @@ def remove_barcodes(reads, config, ntuple=0, verbose=False):
raise ValueError
for adaptator in config:
if (not adaptator == 'conditions' and
- ntuple == config[adaptator]['coords']['reads']-1):
+ ntuple == config[adaptator]['coords']['reads']):
reads = remove_barcode(
reads=reads,
config=config,
@@ -570,7 +570,7 @@ def remove_barcodes(reads, config, ntuple=0, verbose=False):
return reads
-def write_seq(reads, fout, config, ntuple=0, verbose=False):
+def write_seq(reads, fout, config, ntuple, verbose=False):
"""
write sequence without adaptor in the correct file
@@ -617,13 +617,13 @@ def read_reads(fins, reads_list, ntuple, line_number):
"""
if line_number == 0:
for reads in ntuple:
- reads_list[reads-1].header = fins[reads-1]
+ reads_list[reads].header = fins[reads]
if line_number == 1:
for reads in ntuple:
- reads_list[reads-1].seq = fins[reads-1]
+ reads_list[reads].seq = fins[reads]
if line_number == 3:
for reads in ntuple:
- reads_list[reads-1].str2qual(fins[reads-1])
+ reads_list[reads].str2qual(fins[reads])
return reads_list
diff --git a/src/rmi_splitter/tests/data_test.py b/src/rmi_splitter/tests/data_test.py
index f4be63d..447581a 100644
--- a/src/rmi_splitter/tests/data_test.py
+++ b/src/rmi_splitter/tests/data_test.py
@@ -16,7 +16,7 @@ single end
CONFIG_TOY = {
'RT': {
'coords': {
- 'reads': 1,
+ 'reads': 0,
'start': 6,
'stop': 13,
'header': False,
@@ -32,7 +32,7 @@ CONFIG_TOY = {
},
'PCR': {
'coords': {
- 'reads': 1,
+ 'reads': 0,
'start': 0,
'stop': 5,
'header': True,
@@ -49,7 +49,7 @@ CONFIG_TOY = {
},
'UMI': {
'coords': {
- 'reads': 1,
+ 'reads': 0,
'start': 0,
'stop': 5,
'header': False,
@@ -122,7 +122,7 @@ paired-end
CONFIG_TOY_PAIRED = {
'RT': {
'coords': {
- 'reads': 1,
+ 'reads': 0,
'start': 6,
'stop': 13,
'header': False
@@ -136,7 +136,7 @@ CONFIG_TOY_PAIRED = {
},
'PCR': {
'coords': {
- 'reads': 3,
+ 'reads': 2,
'start': 0,
'stop': 5,
'header': False
@@ -152,7 +152,7 @@ CONFIG_TOY_PAIRED = {
},
'UMI': {
'coords': {
- 'reads': 1,
+ 'reads': 0,
'start': 0,
'stop': 5,
'header': False
diff --git a/src/rmi_splitter/tests/rmi_splitter_test.py b/src/rmi_splitter/tests/rmi_splitter_test.py
index aaa9cca..ac2d0be 100644
--- a/src/rmi_splitter/tests/rmi_splitter_test.py
+++ b/src/rmi_splitter/tests/rmi_splitter_test.py
@@ -40,7 +40,7 @@ class ConfigLoadTest(unittest.TestCase):
"""
test on the adaptator position extraction
"""
- pos_object = {'reads': 1,
+ pos_object = {'reads': 0,
'start': 6,
'start_update': 6,
'stop': 13,
@@ -62,7 +62,9 @@ class ConfigLoadTest(unittest.TestCase):
rmi_splitter.list_adaptator_barcode(
config=rmi_splitter.load_yaml(
path="rmi_splitter/tests/data/toy_file.yaml"),
- adaptator="RT"),
+ adaptator="RT",
+ ntuple=0
+ ),
{"RT1": "TAGTGCC",
"RT2": "GCTACCC",
"RT3": "ATCGACC",
@@ -115,7 +117,9 @@ class ReadsReadTest(unittest.TestCase):
}
}
},
- "RT")
+ "RT",
+ 0
+ )
except KeyError:
self.assertEqual(1, 1)
@@ -134,7 +138,9 @@ class ReadsReadTest(unittest.TestCase):
}
}
},
- "RT")
+ "RT",
+ 0
+ )
except KeyError:
self.assertEqual(1, 1)
@@ -153,7 +159,9 @@ class ReadsReadTest(unittest.TestCase):
}
}
},
- "RT")
+ "RT",
+ 0
+ )
except KeyError:
self.assertEqual(1, 1)
@@ -172,7 +180,9 @@ class ReadsReadTest(unittest.TestCase):
}
}
},
- "RT")
+ "RT",
+ 0
+ )
except KeyError:
self.assertEqual(1, 1)
@@ -260,10 +270,12 @@ class ReadsReadTest(unittest.TestCase):
reads_test = copy.deepcopy(data_test.Reads_single)
self.assertEqual(
rmi_splitter.extract_barcode(
- reads_test,
- rmi_splitter.load_yaml(
+ reads=reads_test,
+ config=rmi_splitter.load_yaml(
path="rmi_splitter/tests/data/toy_file.yaml"),
- "RT", verbose=True)['seq'],
+ adaptator="RT",
+ ntuple=0,
+ verbose=True)['seq'],
data_test.RT_barcode_single
)
@@ -274,10 +286,12 @@ class ReadsReadTest(unittest.TestCase):
reads_test = copy.deepcopy(data_test.Reads_single)
self.assertEqual(
rmi_splitter.extract_barcode(
- reads_test,
- rmi_splitter.load_yaml(
+ reads=reads_test,
+ config=rmi_splitter.load_yaml(
path="rmi_splitter/tests/data/toy_file.yaml"),
- "PCR")['seq'],
+ adaptator="PCR",
+ ntuple=0
+ )['seq'],
data_test.PCR_barcode_single
)
@@ -306,7 +320,11 @@ class ModifyReadTest(unittest.TestCase):
reads_test = copy.deepcopy(data_test.Reads_single)
self.assertEqual(
rmi_splitter.write_umi_in_header(
- reads_test, data_test.CONFIG_TOY, 'UMI').header,
+ reads=reads_test,
+ config=data_test.CONFIG_TOY,
+ adaptator='UMI',
+ ntuple=0
+ ).header,
data_test.Reads_umi_single.header
)
@@ -405,9 +423,10 @@ class ModifyReadTest(unittest.TestCase):
reads_test = copy.deepcopy(data_test.Reads_single)
self.assertEqual(
rmi_splitter.remove_barcode(
- reads_test,
- data_test.CONFIG_TOY,
- "RT"
+ reads=reads_test,
+ config=data_test.CONFIG_TOY,
+ adaptator="RT",
+ ntuple=0
).seq,
data_test.Reads_single_noRT.seq
)
@@ -423,7 +442,8 @@ class ModifyReadTest(unittest.TestCase):
self.assertEqual(
rmi_splitter.update_positions(
config=data_test.CONFIG_TOY,
- adaptator="UMI"
+ adaptator="UMI",
+ ntuple=0
),
rmi_splitter.load_yaml(
path="rmi_splitter/tests/data/toy_file.yaml")
@@ -449,15 +469,18 @@ class ModifyReadTest(unittest.TestCase):
reads=reads_test,
config=rmi_splitter.update_positions(
config=data_test.CONFIG_TOY,
- adaptator="UMI"
+ adaptator="UMI",
+ ntuple=0
),
adaptator="PCR",
+ ntuple=0,
barcode_dictionary=barcode_dic
),
rmi_splitter.match_barcode(
reads=reads_test2,
config=data_test.CONFIG_TOY,
adaptator="PCR",
+ ntuple=0,
barcode_dictionary=barcode_dic
),
)
@@ -470,7 +493,8 @@ class ModifyReadTest(unittest.TestCase):
self.assertEqual(
rmi_splitter.remove_barcodes(
reads_test,
- data_test.CONFIG_TOY,
+ ntuple=0,
+ config=data_test.CONFIG_TOY,
).seq,
data_test.Reads_trim_single.seq
)
@@ -495,6 +519,7 @@ class SeachBarcodeTest(unittest.TestCase):
reads=reads_test,
config=config,
adaptator="PCR",
+ ntuple=0,
barcode_dictionary=barcode_dic
),
"PCR1"
@@ -515,6 +540,7 @@ class SeachBarcodeTest(unittest.TestCase):
reads=reads_test,
config=config,
adaptator="PCT",
+ ntuple=0,
barcode_dictionary=barcode_dic,
verbose=True
)
@@ -537,6 +563,7 @@ class SeachBarcodeTest(unittest.TestCase):
rmi_splitter.match_barcodes(
reads=reads_test,
config=config,
+ ntuple=0,
barcode_dictionary=barcode_dic
),
['RT1', 'PCR1']
@@ -584,7 +611,7 @@ class HandleFastqTest(unittest.TestCase):
rmi_splitter.list_reads_number(
config=data_test.CONFIG_TOY_PAIRED
),
- [1, 2, 3]
+ [0, 1, 2]
)
def test_assign_reads_single(self):
@@ -639,7 +666,7 @@ class HandleFastqTest(unittest.TestCase):
config=rmi_splitter.load_yaml(
path="rmi_splitter/tests/data/toy_file_paired.yaml"),
results_path="../results/",
- ntuple_param=range(3),
+ ntuple_param=3,
verbose=True
)
self.assertEqual(1, 1)
--
GitLab