Skip to content

Commit b3b166a

Browse files
drosofffdrosofff
authored
Merge pull request #689 from ARTbio/more_memory
Fix bowtie2-align command !
2 parents 067aa46 + 87acfa9 commit b3b166a

8 files changed

+331
-330
lines changed

‎tools/repenrich2/RepEnrich2.py

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -110,7 +110,7 @@ def run_bowtie(args):
110110
write to files to save memory
111111
'''
112112
metagenome, fastqfile = args
113-
b_opt = "-k 1 -p 1 --quiet --no-hd"
113+
b_opt = "-k 1 -p 1 --quiet --no-hd --no-unal"
114114
command = shlex.split(f"bowtie2 {b_opt} -x {metagenome} {fastqfile}")
115115
bowtie_align = subprocess.run(command, check=True,
116116
capture_output=True, text=True).stdout
@@ -148,6 +148,7 @@ def run_bowtie(args):
148148
repeats_by_reads[read].append(repname)
149149
for repname in repeats_by_reads:
150150
repeats_by_reads[repname] = list(set(repeats_by_reads[repname]))
151+
# this repeats_by_reads dictionary is far too big
151152

152153
# 3 dictionnaries and 1 pointer variable to be populated
153154
fractionalcounts = defaultdict(float)

‎tools/repenrich2/macros.xml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
<macros>
22
<token name="@TOOL_VERSION@">2.31.1</token>
3-
<token name="@VERSION_SUFFIX@">7</token>
3+
<token name="@VERSION_SUFFIX@">8</token>
44
<token name="@PROFILE@">23.0</token>
55

66
<xml name="repenrich_requirements">

‎tools/repenrich2/test-data/chrY_paired_class_fraction_counts.tab

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
1-
DNA 194.66
2-
LINE 279.83
3-
LTR 27051.62
4-
Low_complexity 8.48
5-
RC 4.24
6-
Simple_repeat 161.17
1+
DNA 184.0
2+
LINE 227.0
3+
LTR 27179.0
4+
Low_complexity 0.67
5+
RC 0.0
6+
Simple_repeat 90.33

‎tools/repenrich2/test-data/chrY_paired_family_fraction_counts.tab

Lines changed: 13 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -1,13 +1,13 @@
1-
CMC-Transib 38.48
2-
CR1 8.24
3-
Copia 25859.45
4-
Gypsy 1107.24
5-
Helitron 4.24
6-
Jockey 164.38
7-
LOA 4.24
8-
Low_complexity 8.48
9-
P 40.48
10-
Pao 84.93
11-
R1 102.97
12-
Simple_repeat 161.17
13-
TcMar-Tc1 115.69
1+
CMC-Transib 30.0
2+
CR1 4.0
3+
Copia 25880.0
4+
Gypsy 1238.0
5+
Helitron 0.0
6+
Jockey 107.0
7+
LOA 0.0
8+
Low_complexity 0.67
9+
P 60.0
10+
Pao 61.0
11+
R1 116.0
12+
Simple_repeat 90.33
13+
TcMar-Tc1 94.0

‎tools/repenrich2/test-data/chrY_paired_fraction_counts.tab

Lines changed: 145 additions & 145 deletions
Original file line numberDiff line numberDiff line change
@@ -1,145 +1,145 @@
1-
A-rich Low_complexity Low_complexity 4.24
2-
ACCORD2_I-int LTR Gypsy 4.24
3-
ACCORD2_LTR LTR Gypsy 4.24
4-
ACCORD_I-int LTR Gypsy 4.24
5-
BARI1 DNA TcMar-Tc1 4.24
6-
BATUMI_LTR LTR Pao 4.24
7-
BS LINE Jockey 4.24
8-
BS2 LINE Jockey 62.24
9-
BURDOCK_I-int LTR Gypsy 4.24
10-
Baggins1 LINE LOA 4.24
11-
Bica_I-int LTR Gypsy 42.24
12-
Bica_LTR LTR Gypsy 4.24
13-
CIRCE LTR Gypsy 4.24
14-
Chouto_I-int LTR Gypsy 4.24
15-
Copia1-I_DM LTR Copia 4.24
16-
Copia_I-int LTR Copia 25185.24
17-
Copia_LTR LTR Copia 651.24
18-
DIVER2_I-int LTR Pao 4.24
19-
DIVER2_LTR LTR Pao 4.24
20-
DM1731_I-int LTR Copia 10.24
21-
DM1731_LTR LTR Copia 4.24
22-
DM176_I-int LTR Gypsy 4.24
23-
DM412 LTR Gypsy 4.24
24-
DM412B_LTR LTR Gypsy 4.24
25-
DMCR1A LINE CR1 8.24
26-
DMLTR5 LTR Gypsy 4.24
27-
DMRT1A LINE R1 4.24
28-
DMRT1B LINE R1 90.24
29-
DMRT1C LINE R1 4.24
30-
DNAREP1_DM RC Helitron 4.24
31-
DOC2_DM LINE Jockey 4.24
32-
DOC3_DM LINE Jockey 4.24
33-
FB4_DM DNA TcMar-Tc1 42.24
34-
FROGGER_I-int LTR Copia 4.24
35-
FW2_DM LINE Jockey 4.24
36-
G3_DM LINE Jockey 4.24
37-
G5A_DM LINE Jockey 4.24
38-
G5_DM LINE Jockey 4.24
39-
G6_DM LINE Jockey 4.24
40-
GA-rich Low_complexity Low_complexity 4.24
41-
GTWIN_I-int LTR Gypsy 4.24
42-
G_DM LINE Jockey 4.24
43-
Gypsy11_I-int LTR Gypsy 4.24
44-
Gypsy11_LTR LTR Gypsy 4.24
45-
Gypsy12_LTR LTR Gypsy 4.24
46-
Gypsy2-I_DM LTR Gypsy 6.24
47-
Gypsy2-LTR_DM LTR Gypsy 4.24
48-
Gypsy3_LTR LTR Gypsy 4.24
49-
Gypsy4_I-int LTR Gypsy 4.24
50-
Gypsy5_I-int LTR Gypsy 4.24
51-
Gypsy6A_LTR LTR Gypsy 4.24
52-
Gypsy6_I-int LTR Gypsy 30.24
53-
Gypsy8_I-int LTR Gypsy 4.24
54-
Gypsy8_LTR LTR Gypsy 4.24
55-
Gypsy9_I-int LTR Gypsy 4.24
56-
Gypsy_I-int LTR Gypsy 36.24
57-
Gypsy_LTR LTR Gypsy 5.24
58-
HELENA_RT LINE Jockey 4.24
59-
HETA LINE Jockey 28.24
60-
HMSBEAGLE_I-int LTR Gypsy 6.24
61-
IDEFIX_I-int LTR Gypsy 8.24
62-
IDEFIX_LTR LTR Gypsy 4.24
63-
Invader1_I-int LTR Gypsy 4.24
64-
Invader1_LTR LTR Gypsy 4.24
65-
Invader2_I-int LTR Gypsy 4.24
66-
Invader4_I-int LTR Gypsy 4.24
67-
Invader4_LTR LTR Gypsy 4.24
68-
Invader5_I-int LTR Gypsy 4.24
69-
Invader5_LTR LTR Gypsy 4.24
70-
Invader6_I-int LTR Gypsy 4.24
71-
Invader6_LTR LTR Gypsy 4.24
72-
MAX_I-int LTR Pao 53.24
73-
MAX_LTR LTR Pao 6.24
74-
MDG1_I-int LTR Gypsy 4.24
75-
MDG1_LTR LTR Gypsy 4.24
76-
MDG3_I-int LTR Gypsy 156.24
77-
MDG3_LTR LTR Gypsy 4.24
78-
MICROPIA_I-int LTR Gypsy 4.24
79-
MICROPIA_LTR LTR Gypsy 4.24
80-
Mariner2_DM DNA TcMar-Tc1 4.24
81-
NINJA_I-int LTR Pao 4.24
82-
NOMAD_I-int LTR Gypsy 4.24
83-
PROTOP_A DNA P 36.24
84-
PROTOP_B DNA P 4.24
85-
QUASIMODO2-I_DM LTR Gypsy 46.24
86-
QUASIMODO2-LTR_DM LTR Gypsy 4.24
87-
QUASIMODO_I-int LTR Gypsy 14.24
88-
QUASIMODO_LTR LTR Gypsy 6.24
89-
R1_DM LINE R1 4.24
90-
ROOA_I-int LTR Pao 4.24
91-
ROOA_LTR LTR Pao 4.24
92-
ROVER-I_DM LTR Gypsy 385.24
93-
ROVER-LTR_DM LTR Gypsy 6.24
94-
S2_DM DNA TcMar-Tc1 4.24
95-
STALKER4_I-int LTR Gypsy 81.24
96-
STALKER4_LTR LTR Gypsy 8.24
97-
S_DM DNA TcMar-Tc1 52.24
98-
Stalker2_I-int LTR Gypsy 84.24
99-
Stalker2_LTR LTR Gypsy 6.24
100-
TART-A LINE Jockey 8.24
101-
TART_B1 LINE Jockey 23.24
102-
TC1-2_DM DNA TcMar-Tc1 4.24
103-
TC1_DM DNA TcMar-Tc1 4.24
104-
TLD2 LTR Gypsy 4.24
105-
TRANSIB1 DNA CMC-Transib 4.24
106-
TRANSIB2 DNA CMC-Transib 34.24
107-
ZAM_I-int LTR Gypsy 4.24
108-
_AACACA_n Simple_repeat Simple_repeat 4.24
109-
_AAT_n Simple_repeat Simple_repeat 4.24
110-
_ACAATAG_n Simple_repeat Simple_repeat 4.24
111-
_ACC_n Simple_repeat Simple_repeat 4.24
112-
_AGAGAAG_n Simple_repeat Simple_repeat 4.24
113-
_AGAGA_n Simple_repeat Simple_repeat 4.24
114-
_ATAAT_n Simple_repeat Simple_repeat 4.24
115-
_ATATATT_n Simple_repeat Simple_repeat 4.24
116-
_ATATTAT_n Simple_repeat Simple_repeat 4.24
117-
_ATTTTT_n Simple_repeat Simple_repeat 4.24
118-
_ATT_n Simple_repeat Simple_repeat 4.24
119-
_AT_n Simple_repeat Simple_repeat 4.24
120-
_A_n Simple_repeat Simple_repeat 4.24
121-
_CATA_n Simple_repeat Simple_repeat 4.24
122-
_CTTTT_n Simple_repeat Simple_repeat 4.24
123-
_GAGAA_n Simple_repeat Simple_repeat 4.24
124-
_GCCTTT_n Simple_repeat Simple_repeat 4.24
125-
_TAATAT_n Simple_repeat Simple_repeat 4.24
126-
_TAATA_n Simple_repeat Simple_repeat 4.24
127-
_TATAAAA_n Simple_repeat Simple_repeat 4.24
128-
_TATAA_n Simple_repeat Simple_repeat 4.24
129-
_TATCATG_n Simple_repeat Simple_repeat 4.24
130-
_TA_n Simple_repeat Simple_repeat 4.24
131-
_TGTTG_n Simple_repeat Simple_repeat 4.24
132-
_TTATATA_n Simple_repeat Simple_repeat 4.24
133-
_TTATAT_n Simple_repeat Simple_repeat 4.24
134-
_TTATA_n Simple_repeat Simple_repeat 4.24
135-
_TTA_n Simple_repeat Simple_repeat 4.24
136-
_TTCTT_n Simple_repeat Simple_repeat 4.24
137-
_TTC_n Simple_repeat Simple_repeat 4.24
138-
_TTTAT_n Simple_repeat Simple_repeat 4.24
139-
_TTTA_n Simple_repeat Simple_repeat 4.24
140-
_TTTC_n Simple_repeat Simple_repeat 4.24
141-
_TTTGA_n Simple_repeat Simple_repeat 4.24
142-
_TTTTAG_n Simple_repeat Simple_repeat 4.24
143-
_TTTTCTT_n Simple_repeat Simple_repeat 4.24
144-
_TTTTC_n Simple_repeat Simple_repeat 4.24
145-
_T_n Simple_repeat Simple_repeat 4.24
1+
A-rich Low_complexity Low_complexity 0.0
2+
ACCORD2_I-int LTR Gypsy 0.0
3+
ACCORD2_LTR LTR Gypsy 0.0
4+
ACCORD_I-int LTR Gypsy 0.0
5+
BARI1 DNA TcMar-Tc1 0.0
6+
BATUMI_LTR LTR Pao 0.0
7+
BS LINE Jockey 0.0
8+
BS2 LINE Jockey 58.0
9+
BURDOCK_I-int LTR Gypsy 0.0
10+
Baggins1 LINE LOA 0.0
11+
Bica_I-int LTR Gypsy 49.0
12+
Bica_LTR LTR Gypsy 1.0
13+
CIRCE LTR Gypsy 0.0
14+
Chouto_I-int LTR Gypsy 1.5
15+
Copia1-I_DM LTR Copia 0.0
16+
Copia_I-int LTR Copia 25220.0
17+
Copia_LTR LTR Copia 652.0
18+
DIVER2_I-int LTR Pao 1.0
19+
DIVER2_LTR LTR Pao 2.0
20+
DM1731_I-int LTR Copia 8.0
21+
DM1731_LTR LTR Copia 0.0
22+
DM176_I-int LTR Gypsy 0.0
23+
DM412 LTR Gypsy 12.5
24+
DM412B_LTR LTR Gypsy 0.0
25+
DMCR1A LINE CR1 4.0
26+
DMLTR5 LTR Gypsy 0.0
27+
DMRT1A LINE R1 0.0
28+
DMRT1B LINE R1 116.0
29+
DMRT1C LINE R1 0.0
30+
DNAREP1_DM RC Helitron 0.0
31+
DOC2_DM LINE Jockey 0.0
32+
DOC3_DM LINE Jockey 0.0
33+
FB4_DM DNA TcMar-Tc1 41.0
34+
FROGGER_I-int LTR Copia 0.0
35+
FW2_DM LINE Jockey 0.0
36+
G3_DM LINE Jockey 0.0
37+
G5A_DM LINE Jockey 0.0
38+
G5_DM LINE Jockey 0.0
39+
G6_DM LINE Jockey 0.0
40+
GA-rich Low_complexity Low_complexity 0.67
41+
GTWIN_I-int LTR Gypsy 12.5
42+
G_DM LINE Jockey 0.0
43+
Gypsy11_I-int LTR Gypsy 0.0
44+
Gypsy11_LTR LTR Gypsy 0.0
45+
Gypsy12_LTR LTR Gypsy 0.0
46+
Gypsy2-I_DM LTR Gypsy 4.0
47+
Gypsy2-LTR_DM LTR Gypsy 0.0
48+
Gypsy3_LTR LTR Gypsy 0.0
49+
Gypsy4_I-int LTR Gypsy 0.0
50+
Gypsy5_I-int LTR Gypsy 0.0
51+
Gypsy6A_LTR LTR Gypsy 1.0
52+
Gypsy6_I-int LTR Gypsy 31.0
53+
Gypsy8_I-int LTR Gypsy 0.0
54+
Gypsy8_LTR LTR Gypsy 0.0
55+
Gypsy9_I-int LTR Gypsy 0.0
56+
Gypsy_I-int LTR Gypsy 32.0
57+
Gypsy_LTR LTR Gypsy 1.0
58+
HELENA_RT LINE Jockey 0.0
59+
HETA LINE Jockey 24.0
60+
HMSBEAGLE_I-int LTR Gypsy 2.0
61+
IDEFIX_I-int LTR Gypsy 8.0
62+
IDEFIX_LTR LTR Gypsy 2.0
63+
Invader1_I-int LTR Gypsy 0.0
64+
Invader1_LTR LTR Gypsy 0.0
65+
Invader2_I-int LTR Gypsy 0.0
66+
Invader4_I-int LTR Gypsy 0.0
67+
Invader4_LTR LTR Gypsy 0.0
68+
Invader5_I-int LTR Gypsy 0.0
69+
Invader5_LTR LTR Gypsy 0.0
70+
Invader6_I-int LTR Gypsy 0.0
71+
Invader6_LTR LTR Gypsy 0.0
72+
MAX_I-int LTR Pao 56.0
73+
MAX_LTR LTR Pao 2.0
74+
MDG1_I-int LTR Gypsy 0.0
75+
MDG1_LTR LTR Gypsy 0.0
76+
MDG3_I-int LTR Gypsy 156.5
77+
MDG3_LTR LTR Gypsy 2.5
78+
MICROPIA_I-int LTR Gypsy 51.0
79+
MICROPIA_LTR LTR Gypsy 2.0
80+
Mariner2_DM DNA TcMar-Tc1 0.0
81+
NINJA_I-int LTR Pao 0.0
82+
NOMAD_I-int LTR Gypsy 0.0
83+
PROTOP_A DNA P 50.0
84+
PROTOP_B DNA P 10.0
85+
QUASIMODO2-I_DM LTR Gypsy 43.0
86+
QUASIMODO2-LTR_DM LTR Gypsy 0.0
87+
QUASIMODO_I-int LTR Gypsy 108.0
88+
QUASIMODO_LTR LTR Gypsy 23.0
89+
R1_DM LINE R1 0.0
90+
ROOA_I-int LTR Pao 0.0
91+
ROOA_LTR LTR Pao 0.0
92+
ROVER-I_DM LTR Gypsy 414.0
93+
ROVER-LTR_DM LTR Gypsy 6.0
94+
S2_DM DNA TcMar-Tc1 0.0
95+
STALKER4_I-int LTR Gypsy 143.5
96+
STALKER4_LTR LTR Gypsy 25.0
97+
S_DM DNA TcMar-Tc1 53.0
98+
Stalker2_I-int LTR Gypsy 103.0
99+
Stalker2_LTR LTR Gypsy 3.0
100+
TART-A LINE Jockey 4.0
101+
TART_B1 LINE Jockey 21.0
102+
TC1-2_DM DNA TcMar-Tc1 0.0
103+
TC1_DM DNA TcMar-Tc1 0.0
104+
TLD2 LTR Gypsy 0.0
105+
TRANSIB1 DNA CMC-Transib 0.0
106+
TRANSIB2 DNA CMC-Transib 30.0
107+
ZAM_I-int LTR Gypsy 0.0
108+
_AACACA_n Simple_repeat Simple_repeat 0.0
109+
_AAT_n Simple_repeat Simple_repeat 0.0
110+
_ACAATAG_n Simple_repeat Simple_repeat 0.0
111+
_ACC_n Simple_repeat Simple_repeat 0.0
112+
_AGAGAAG_n Simple_repeat Simple_repeat 2.17
113+
_AGAGA_n Simple_repeat Simple_repeat 43.0
114+
_ATAAT_n Simple_repeat Simple_repeat 0.0
115+
_ATATATT_n Simple_repeat Simple_repeat 0.0
116+
_ATATTAT_n Simple_repeat Simple_repeat 0.0
117+
_ATTTTT_n Simple_repeat Simple_repeat 0.0
118+
_ATT_n Simple_repeat Simple_repeat 0.0
119+
_AT_n Simple_repeat Simple_repeat 0.0
120+
_A_n Simple_repeat Simple_repeat 0.0
121+
_CATA_n Simple_repeat Simple_repeat 0.0
122+
_CTTTT_n Simple_repeat Simple_repeat 0.0
123+
_GAGAA_n Simple_repeat Simple_repeat 45.17
124+
_GCCTTT_n Simple_repeat Simple_repeat 0.0
125+
_TAATAT_n Simple_repeat Simple_repeat 0.0
126+
_TAATA_n Simple_repeat Simple_repeat 0.0
127+
_TATAAAA_n Simple_repeat Simple_repeat 0.0
128+
_TATAA_n Simple_repeat Simple_repeat 0.0
129+
_TATCATG_n Simple_repeat Simple_repeat 0.0
130+
_TA_n Simple_repeat Simple_repeat 0.0
131+
_TGTTG_n Simple_repeat Simple_repeat 0.0
132+
_TTATATA_n Simple_repeat Simple_repeat 0.0
133+
_TTATAT_n Simple_repeat Simple_repeat 0.0
134+
_TTATA_n Simple_repeat Simple_repeat 0.0
135+
_TTA_n Simple_repeat Simple_repeat 0.0
136+
_TTCTT_n Simple_repeat Simple_repeat 0.0
137+
_TTC_n Simple_repeat Simple_repeat 0.0
138+
_TTTAT_n Simple_repeat Simple_repeat 0.0
139+
_TTTA_n Simple_repeat Simple_repeat 0.0
140+
_TTTC_n Simple_repeat Simple_repeat 0.0
141+
_TTTGA_n Simple_repeat Simple_repeat 0.0
142+
_TTTTAG_n Simple_repeat Simple_repeat 0.0
143+
_TTTTCTT_n Simple_repeat Simple_repeat 0.0
144+
_TTTTC_n Simple_repeat Simple_repeat 0.0
145+
_T_n Simple_repeat Simple_repeat 0.0

‎tools/repenrich2/test-data/chrY_single_class_fraction_counts.tab

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
1-
DNA 121.89
2-
LINE 184.34
3-
LTR 13554.36
4-
Low_complexity 9.43
5-
RC 4.72
6-
Simple_repeat 179.26
1+
DNA 109.0
2+
LINE 129.0
3+
LTR 13705.0
4+
Low_complexity 0.0
5+
RC 0.0
6+
Simple_repeat 91.0

‎tools/repenrich2/test-data/chrY_single_family_fraction_counts.tab

Lines changed: 13 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -1,13 +1,13 @@
1-
CMC-Transib 21.43
2-
CR1 6.72
3-
Copia 12817.3
4-
Gypsy 670.32
5-
Helitron 4.72
6-
Jockey 119.04
7-
LOA 4.72
8-
Low_complexity 9.43
9-
P 27.43
10-
Pao 66.74
11-
R1 53.87
12-
Simple_repeat 179.26
13-
TcMar-Tc1 73.02
1+
CMC-Transib 15.0
2+
CR1 2.0
3+
Copia 12865.0
4+
Gypsy 805.0
5+
Helitron 0.0
6+
Jockey 54.0
7+
LOA 0.0
8+
Low_complexity 0.0
9+
P 44.0
10+
Pao 35.0
11+
R1 73.0
12+
Simple_repeat 91.0
13+
TcMar-Tc1 50.0

0 commit comments

Comments
 (0)