Mercurial > repos > iuc > tooldistillator

changeset 5:66617be450bd draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/tooldistillator commit 164bcc563f8495d4bb1ce14b574aef9287904f40
author iuc
date Thu, 10 Apr 2025 15:02:10 +0000
parents ea93df4b3df2
children dfd6fa2637ba
files macro.xml test-data/checkm2/DIAMOND_RESULTS.tsv test-data/checkm2/checkm2.log test-data/checkm2/quality_report.tsv tooldistillator.xml
diffstat 5 files changed, 219 insertions(+), 19 deletions(-) [+]
line wrap: on
line diff
--- a/macro.xml	Thu Oct 17 08:00:15 2024 +0000
+++ b/macro.xml	Thu Apr 10 15:02:10 2025 +0000
@@ -1,7 +1,7 @@
 <?xml version="1.0"?>
 <macros>
-    <token name="@TOOL_VERSION@">0.9.1</token>
-    <token name="@VERSION_SUFFIX@">1</token>
+    <token name="@TOOL_VERSION@">0.9.2</token>
+    <token name="@VERSION_SUFFIX@">0</token>
     <token name="@PROFILE@">21.05</token>
     <xml name="version_command">
         <version_command><![CDATA[tooldistillator --version]]></version_command>
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/checkm2/DIAMOND_RESULTS.tsv	Thu Apr 10 15:02:10 2025 +0000
@@ -0,0 +1,100 @@
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49428_2	UniRef100_A0A0N7IEW1~K00616	91.7	218	18	0	1	218	1	218	3.7e-113	412.5
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49428_3	UniRef100_B5R745~K05776	39.4	358	206	4	14	367	49	399	3.5e-63	247.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_28292_2	UniRef100_E6STV5~K03402	94.3	157	9	0	1	157	1	157	3.4e-76	289.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_28292_4	UniRef100_A0A1A9I3U8~K05970	45.9	427	197	7	12	409	12	433	1.2e-101	375.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_14200_2	UniRef100_E6SQC7~K01200	86.9	665	87	0	1	665	1	665	0.0e+00	1218.0
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_63629_1	UniRef100_E6SR16~K03711	95.1	162	7	1	1	161	1	162	6.6e-83	311.6
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_114_1	UniRef100_A0A139LU84~K01952	94.2	1011	59	0	1	1011	38	1048	0.0e+00	1941.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_21313_3	UniRef100_E6SMX3~K06950	78.4	208	45	0	1	208	1	208	3.4e-92	342.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_21313_4	UniRef100_E6SMX8~K06942	94.6	367	20	0	1	367	1	367	3.7e-198	695.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_21322_2	UniRef100_A0A0P0G6A0~K05349	80.0	862	172	0	1	862	2	863	0.0e+00	1431.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7210_2	UniRef100_A0A0P0GLB2~K14742	92.2	230	18	0	1	230	1	230	8.7e-121	438.0
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7210_4	UniRef100_E6SRP6~K00942	90.9	187	17	0	1	187	1	187	1.5e-91	340.5
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7222_2	UniRef100_A0A127AYQ9~K21573	33.3	1066	648	24	3	1045	1	1026	8.1e-142	510.0
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7222_5	UniRef100_S4XS10~K15921	31.9	467	248	23	20	442	84	524	4.9e-48	197.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_63747_1	UniRef100_A0A1T5DC71~K02343	70.7	362	106	0	4	365	3	364	6.5e-146	522.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_14378_2	UniRef100_Q7UNN1~K01134	43.8	491	265	6	28	512	40	525	6.7e-113	412.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_42512_3	UniRef100_A3U9J4~K08316	38.8	165	97	3	2	164	16	178	5.3e-27	125.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_42512_4	UniRef100_T1YCJ1~K06131	44.5	272	148	3	1	269	33	304	6.9e-61	239.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_14415_3	UniRef100_E6STG4~K02619	68.2	201	58	1	1	201	1	195	1.1e-71	274.6
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_14415_4	UniRef100_X2K1F6~K01665	56.5	248	108	0	5	252	30	277	4.2e-76	289.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_28506_2	UniRef100_E6SNZ6~K12573	92.3	711	55	0	4	714	7	717	0.0e+00	1304.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_28506_3	UniRef100_I9Q1S0~K07005	74.1	158	41	0	1	158	1	158	1.0e-67	261.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_35544_1	UniRef100_E6SPZ0~K00873	94.5	458	25	0	1	458	28	485	2.6e-246	855.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_35544_2	UniRef100_C7M9L1~K00588	44.0	207	115	1	3	209	8	213	8.0e-49	198.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_247_2	UniRef100_A0A089KNB8~K01198&K01209	31.1	492	280	21	27	487	10	473	1.3e-39	169.5
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_247_3	UniRef100_A0A0P0GKF6~K15923	70.2	1071	312	4	12	1080	8	1073	0.0e+00	1597.4
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49649_3	UniRef100_A0A0K1KXX7~K07658	39.6	230	128	3	2	222	3	230	9.9e-37	158.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49649_5	UniRef100_A0A0P0G8E4~K06142	88.1	201	23	1	1	200	1	201	1.5e-89	334.0
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49664_3	UniRef100_A0A0E3PP05~K00763	38.9	265	150	4	1	254	1	264	1.2e-42	178.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_63921_2	UniRef100_E6SQK9~K20866	79.0	205	43	0	1	205	1	205	1.5e-92	344.0
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_63921_3	UniRef100_A0A0P0G8Y8~K01537	89.8	842	86	0	1	842	53	894	0.0e+00	1457.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_63924_3	UniRef100_A0A290SA98~K15726	37.3	381	237	2	1	381	44	422	1.4e-70	271.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49669_2	UniRef100_A0A127AYQ9~K21573	35.7	1080	585	33	13	1040	4	1026	2.7e-161	574.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49669_3	UniRef100_A0A0P0GBJ8~K21572	69.2	539	152	5	1	528	1	536	9.6e-216	754.6
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49669_6	UniRef100_A0A0P0GI09~K01190	84.9	1050	159	0	1	1050	1	1050	0.0e+00	1939.1
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7445_2	UniRef100_B5YB77~K15921	35.2	474	247	19	26	453	48	507	9.0e-61	239.6
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7453_2	UniRef100_E6SR03~K05516	85.8	317	40	2	1	317	1	312	5.0e-151	538.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_35703_1	UniRef100_U5Q4K6~K03183	41.8	153	87	1	9	159	1	153	1.9e-29	134.0
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_35703_2	UniRef100_E6SV36~K00014	92.7	248	18	0	1	248	1	248	3.7e-133	479.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_35703_3	UniRef100_A0A0P0G426~K06889	78.7	315	67	0	1	315	1	315	1.3e-148	530.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_35703_4	UniRef100_A0A0P0GMY7~K06872	81.4	242	45	0	5	246	1	242	2.9e-114	416.4
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_56776_2	UniRef100_A0A175ACE0~K05297	56.0	50	22	0	3	52	2	51	2.0e-12	75.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_56776_3	UniRef100_A0A0P0GR35~K03324	93.6	565	36	0	1	565	1	565	1.1e-294	1016.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_56776_4	UniRef100_B9KAM4~K00876	40.3	370	208	5	21	384	29	391	1.6e-66	258.5
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_427_1	UniRef100_F0YI08~K00134	73.4	79	21	0	7	85	6	84	9.2e-26	120.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_427_2	UniRef100_E6SV71~K01284	81.9	685	124	0	1	685	1	685	0.0e+00	1172.1
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_28766_1	UniRef100_A0A140NMH2~K00262	60.7	84	32	1	1	84	17	99	1.9e-20	103.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_28766_3	UniRef100_E6SRR4~K16089	72.3	750	185	5	3	749	6	735	0.0e+00	1114.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_28848_1	UniRef100_I1YI16~K07689	31.3	208	141	2	6	212	6	212	1.9e-21	107.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7592_1	UniRef100_E6SR99~K07037	87.6	687	84	1	1	687	1	686	0.0e+00	1193.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7592_3	UniRef100_R6LHP4~K01104	86.6	149	20	0	6	154	33	181	1.6e-73	280.4
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7592_4	UniRef100_B3DWK1~K04066	40.8	534	283	10	14	527	110	630	4.6e-109	400.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_21835_1	UniRef100_E6SRJ9~K03702	91.4	685	58	1	1	685	1	684	0.0e+00	1232.6
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_21835_2	UniRef100_E6SRK0~K01912	93.5	432	28	0	1	432	1	432	1.1e-241	840.5
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_21835_4	UniRef100_A0A0P0GAI4~K05807	93.3	267	18	0	1	267	1	267	4.5e-137	492.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_57056_1	UniRef100_E6ST38~K00053	95.7	347	15	0	2	348	1	347	2.0e-190	669.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_57056_2	UniRef100_B0D4X2~K19783	31.1	591	312	19	112	621	127	703	9.9e-50	203.4
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49958_2	UniRef100_E6SQ13~K01243	83.3	234	39	0	1	234	1	234	1.2e-109	401.0
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49958_3	UniRef100_E6SQ14~K07173	93.2	162	11	0	1	162	1	162	2.9e-86	322.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49958_4	UniRef100_R4M9B3~K00666	49.3	294	146	3	5	295	8	301	1.3e-76	291.6
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_810_1	UniRef100_A0A0P0GIW9~K03657	87.2	678	86	1	2	679	3	679	0.0e+00	1186.4
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49992_1	UniRef100_A0A0N7IFG4~K00789	94.5	419	23	0	1	419	12	430	2.8e-226	789.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49992_2	UniRef100_A0A0P0FY07~K06966	83.1	195	33	0	2	196	1	195	7.2e-92	341.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49992_4	UniRef100_E6SNW4~K01719	95.6	250	11	0	1	250	1	250	3.3e-134	482.6
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_49992_5	UniRef100_E6SNW5~K03536	85.2	108	16	0	1	108	1	108	9.9e-43	177.6
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_36157_2	UniRef100_E6SUR0~K00948	96.8	312	10	0	1	312	1	312	7.0e-166	588.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_14711_1	UniRef100_E6SNY0~K09815	72.0	304	80	1	11	309	1	304	3.0e-124	449.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_14711_2	UniRef100_A0A0P0GQ72~K09817	85.0	254	33	1	1	254	1	249	6.9e-119	431.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7946_1	UniRef100_Q5LHW2~K02884	99.1	117	1	0	1	117	1	117	1.5e-57	226.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7946_2	UniRef100_A0A0N7IF60~K00845	94.8	326	17	0	1	326	1	326	8.6e-175	617.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7946_3	UniRef100_A0A0P0FVB4~K02003	95.0	238	12	0	1	238	1	238	4.2e-126	455.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7946_4	UniRef100_E6STT3~K02004	83.3	419	69	1	1	418	1	419	9.6e-203	711.1
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7946_5	UniRef100_E6STT4~K02004	88.9	413	46	0	1	413	1	413	1.7e-212	743.4
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7946_6	UniRef100_E6STT5~K02005	91.3	368	32	0	1	368	1	368	2.2e-179	633.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7946_7	UniRef100_E6STT6~K12340	85.2	433	63	1	3	434	10	442	2.2e-202	709.9
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7946_8	UniRef100_E6STU0~K00243	84.9	278	42	0	1	278	1	278	2.4e-136	490.0
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_7946_9	UniRef100_E6STU7~K00024	94.0	333	20	0	1	333	1	333	7.5e-182	641.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_1015_1	UniRef100_A0A1U7XXK2~K00895	52.9	272	126	1	4	273	31	302	3.4e-74	283.5
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_1015_2	UniRef100_A0A174D557~K18139	75.0	456	114	0	1	456	1	456	2.2e-192	676.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_1015_3	UniRef100_T0GCG4~K18138	39.4	589	339	8	5	579	12	596	7.1e-103	379.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_22153_3	UniRef100_E6STN2~K02069	92.8	263	19	0	1	263	1	263	5.9e-129	465.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_22153_5	UniRef100_E6STN3~K02068	79.2	207	43	0	1	207	1	207	5.5e-90	335.5
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_57444_2	UniRef100_A0A0N7IFV5~K03118	90.9	263	24	0	17	279	1	263	9.6e-133	478.0
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_57444_3	UniRef100_D4VLF0~K03116	95.8	72	3	0	3	74	1	72	1.7e-17	93.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_57444_4	UniRef100_B6YQB3~K01775	50.8	311	141	2	1	299	70	380	2.0e-80	304.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_1116_3	UniRef100_E6SQF3~K01224	73.2	392	103	1	3	394	5	394	3.0e-174	616.3
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_50177_4	UniRef100_A0A1L3GGF5~K11383	36.4	236	148	2	20	255	1	234	1.6e-35	154.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_36310_1	UniRef100_A0A076H3R8~K15633	41.6	341	184	5	1	330	34	370	8.7e-66	255.8
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_36310_3	UniRef100_L0RWC6~K02470	50.9	273	125	5	25	296	8	272	9.9e-69	265.4
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_14955_2	UniRef100_Q7MX03~K01070	63.5	244	89	0	29	272	6	249	2.1e-94	350.5
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_14955_3	UniRef100_A0A0P0G3B8~K03699	87.0	430	56	0	1	430	10	439	1.3e-202	710.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_15011_4	UniRef100_A0A0P0GFV2~K01735	90.1	353	35	0	1	353	1	353	2.4e-186	656.4
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_15043_2	UniRef100_A0A0P0GCU8~K07568	95.5	352	16	0	1	352	1	352	1.5e-193	680.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_15043_3	UniRef100_A0A0P0M190~K00950	74.3	148	38	0	1	148	5	152	3.0e-58	229.6
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_15043_4	UniRef100_A0A076IYN8~K21557	33.1	523	339	5	18	538	18	531	3.4e-75	287.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_1350_6	UniRef100_H6SQ71~K03701	44.9	510	274	4	5	508	17	525	1.4e-110	405.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_15093_1	UniRef100_D4IXW5~K04485	48.2	141	66	3	1	138	1	137	2.8e-30	136.7
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_15093_2	UniRef100_B6VS51~K01673	83.8	179	29	0	1	179	3	181	2.5e-83	313.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_15093_3	UniRef100_E6SSR0~K01424	87.5	345	43	0	1	345	1	345	1.8e-175	620.2
+bin_semibin_SASB_CSM7KOLE.1ΩCSM7KOLE_k117_15116_2	UniRef100_A0A0P0GBM8~K01209	87.9	514	62	0	1	514	1	514	1.7e-278	963.0
\ No newline at end of file
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/checkm2/checkm2.log	Thu Apr 10 15:02:10 2025 +0000
@@ -0,0 +1,11 @@
+[03/03/2025 11:20:03 AM] INFO: Running CheckM2 version 1.0.2
+[03/03/2025 11:20:03 AM] INFO: Custom database path provided for predict run. Checking database at Database/CheckM2_database/uniref100.KO.1.dmnd...
+[03/03/2025 11:20:10 AM] INFO: Running quality prediction workflow with 5 threads.
+[03/03/2025 11:20:12 AM] INFO: Calling genes in 50 bins with 5 threads:
+[03/03/2025 11:20:59 AM] WARNING: Skipping protein file batches/batch_0/checkm2/protein_files/bin_concoct_SASB_HSMA33OT.11.faa as it was empty.
+[03/03/2025 11:20:59 AM] INFO: Calculating metadata for 49 bins with 5 threads:
+[03/03/2025 11:20:59 AM] INFO: Annotating input genomes with DIAMOND using 5 threads
+[03/03/2025 11:23:40 AM] INFO: Processing DIAMOND output
+[03/03/2025 11:23:40 AM] INFO: Predicting completeness and contamination using ML models.
+[03/03/2025 11:23:44 AM] INFO: Parsing all results and constructing final output table.
+[03/03/2025 11:23:44 AM] INFO: CheckM2 finished successfully.
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/checkm2/quality_report.tsv	Thu Apr 10 15:02:10 2025 +0000
@@ -0,0 +1,50 @@
+Name	Completeness	Contamination	Completeness_Model_Used	Translation_Table_Used	Coding_Density	Contig_N50	Average_Gene_Length	Genome_Size	GC_Content	Total_Coding_Sequences	Total_Contigs	Max_Contig_Length	Additional_Notes
+bin_concoct_SASB_CSM7KOLE.0	22.51	0.0	Neural Network (Specific Model)	11	0.887	2144	267.22576530612247	708085	0.53	784	324	5294	None
+bin_concoct_SASB_CSM7KOLE.1	15.91	5.09	Neural Network (Specific Model)	11	0.845	4814	222.07713884992987	1119464	0.47	1426	273	50470	None
+bin_concoct_SASB_CSM7KOLE.20	6.38	0.0	Gradient Boost (General Model)	11	0.974	3001	300.0	16618	0.61	18	6	4049	None
+bin_concoct_SASB_CSM7KOLE.22	100.0	2.05	Neural Network (Specific Model)	11	0.87	22717	300.98043641835966	2755022	0.54	2658	187	146684	None
+bin_concoct_SASB_CSM7KOLE.24	6.38	0.0	Gradient Boost (General Model)	11	0.926	5194	268.5	5194	0.45	6	1	5194	None
+bin_concoct_SASB_CSM7KOLE.3	6.38	0.0	Gradient Boost (General Model)	11	0.818	4650	264.75	7724	0.36	8	3	4650	None
+bin_concoct_SASB_CSM7KOLE.30	6.38	0.0	Gradient Boost (General Model)	11	0.796	1887	501.0	1887	0.5	1	1	1887	None
+bin_concoct_SASB_HSMA33MG.1	33.58	0.68	Neural Network (Specific Model)	11	0.856	2959	250.53061224489795	987394	0.59	1127	345	13858	None
+bin_concoct_SASB_HSMA33MG.10	6.38	0.0	Gradient Boost (General Model)	11	0.998	2318	386.0	2318	0.36	2	1	2318	None
+bin_concoct_SASB_HSMA33MG.12	9.66	0.29	Neural Network (Specific Model)	11	0.887	1956	238.57806691449815	432765	0.41	538	204	12095	None
+bin_concoct_SASB_HSMA33MG.13	6.65	0.0	Gradient Boost (General Model)	11	0.936	1726	269.5	1726	0.45	2	1	1726	None
+bin_concoct_SASB_HSMA33MG.8	6.38	0.0	Gradient Boost (General Model)	11	0.794	1759	233.0	1759	0.53	2	1	1759	None
+bin_concoct_SASB_HSMA33MG.9	53.69	21.09	Neural Network (Specific Model)	11	0.901	2957	269.85174585002864	3130308	0.44	3494	1086	18806	None
+bin_concoct_SASB_HSMA33OT.0	4.43	0.0	Neural Network (Specific Model)	11	0.851	1985	214.93157894736842	143344	0.38	190	71	3816	None
+bin_concoct_SASB_HSMA33OT.10	6.38	0.0	Gradient Boost (General Model)	11	0.966	11244	210.76666666666668	19477	0.36	30	4	11244	None
+bin_concoct_SASB_HSMA33OT.14	6.38	0.0	Gradient Boost (General Model)	11	0.509	1707	290.0	1707	0.44	1	1	1707	None
+bin_concoct_SASB_HSMA33OT.16	9.03	0.06	Neural Network (Specific Model)	11	0.881	1824	231.57877813504822	244747	0.42	311	130	3723	None
+bin_concoct_SASB_HSMA33OT.18	6.38	0.0	Gradient Boost (General Model)	11	0.958	1602	256.0	1602	0.36	2	1	1602	None
+bin_concoct_SASB_HSMA33OT.2	6.38	0.0	Gradient Boost (General Model)	11	0.941	1704	345.0	3299	0.35	3	2	1704	None
+bin_concoct_SASB_HSMA33OT.3	5.64	0.0	Gradient Boost (General Model)	11	0.839	2532	203.5	4366	0.44	6	2	2532	None
+bin_concoct_SASB_HSMA33OT.4	6.38	0.0	Gradient Boost (General Model)	11	0.634	2866	202.0	2866	0.42	3	1	2866	None
+bin_concoct_SASB_HSMA33OT.6	6.42	0.0	Gradient Boost (General Model)	11	0.915	1816	233.42857142857142	10707	0.61	14	6	2097	None
+bin_concoct_SASB_MSM5LLDE.16	6.38	0.0	Gradient Boost (General Model)	11	0.998	1592	530.0	1592	0.6	1	1	1592	None
+bin_concoct_SASB_MSM5LLDE.19	6.38	0.0	Gradient Boost (General Model)	11	0.819	1941	68.75	1941	0.39	8	1	1941	None
+bin_concoct_SASB_MSM5LLDE.2	6.38	0.0	Gradient Boost (General Model)	11	0.876	1618	236.5	1618	0.54	2	1	1618	None
+bin_concoct_SASB_MSM5LLDE.20	49.54	0.72	Neural Network (Specific Model)	11	0.901	2559	261.2743158588856	2632930	0.42	3033	1029	33333	None
+bin_concoct_SASB_MSM5LLDE.21	99.55	8.64	Neural Network (Specific Model)	11	0.9	81127	346.90618336886996	4332951	0.46	3752	143	235469	None
+bin_concoct_SASB_MSM5LLDE.27	9.51	0.07	Neural Network (Specific Model)	11	0.877	1958	217.07469879518072	307709	0.57	415	150	5816	None
+bin_concoct_SASB_MSM5LLDE.28	6.38	0.0	Gradient Boost (General Model)	11	0.952	1638	260.0	1638	0.59	2	1	1638	None
+bin_concoct_SASB_MSM5LLDE.6	5.91	0.0	Gradient Boost (General Model)	11	0.91	1987	228.10714285714286	20991	0.61	28	11	2365	None
+bin_concoct_SASB_MSM5LLDE.7	5.03	0.0	Gradient Boost (General Model)	11	0.999	1689	282.0	1689	0.65	2	1	1689	None
+bin_metabat_SASB_CSM7KOLE.10	45.04	8.76	Neural Network (Specific Model)	11	0.903	2382	261.277639046538	3814277	0.41	4405	1544	17510	None
+bin_metabat_SASB_CSM7KOLE.2	18.36	0.04	Neural Network (Specific Model)	11	0.924	2177	272.32474226804123	514039	0.49	582	230	4454	None
+bin_metabat_SASB_CSM7KOLE.3	15.42	0.06	Neural Network (Specific Model)	11	0.944	2547	288.0351906158358	623088	0.49	682	253	7390	None
+bin_metabat_SASB_CSM7KOLE.7	5.83	0.18	Neural Network (Specific Model)	11	0.906	146684	275.6360153256705	237533	0.52	261	10	146684	None
+bin_metabat_SASB_MSM5LLDE.1	85.75	0.14	Neural Network (Specific Model)	11	0.897	81302	343.6424261670372	3612719	0.46	3149	65	235469	None
+bin_metabat_SASB_MSM5LLDE.2	96.34	0.33	Neural Network (Specific Model)	11	0.866	76315	307.9110091743119	2321808	0.57	2180	40	221393	None
+bin_semibin_SASB_CSM7KOLE.0	17.68	0.13	Neural Network (Specific Model)	11	0.895	3290	276.360396039604	467082	0.59	505	137	14331	None
+bin_semibin_SASB_CSM7KOLE.1	45.77	0.88	Gradient Boost (General Model)	11	0.901	4330	318.24455825864277	1653306	0.5	1562	386	16396	None
+bin_semibin_SASB_CSM7KOLE.10	10.46	0.46	Neural Network (Specific Model)	11	0.884	8223	264.5268817204301	749398	0.39	837	118	30972	None
+bin_semibin_SASB_CSM7KOLE.11	14.55	3.97	Neural Network (Specific Model)	11	0.89	19781	259.4004424778761	788274	0.5	904	74	146684	None
+bin_semibin_SASB_CSM7KOLE.14	92.78	0.09	Neural Network (Specific Model)	11	0.897	65685	352.9744467860906	4474676	0.42	3796	113	353143	None
+bin_semibin_SASB_CSM7KOLE.16	10.61	0.0	Neural Network (Specific Model)	11	0.881	3151	281.17182130584195	278023	0.52	291	84	5610	None
+bin_semibin_SASB_CSM7KOLE.3	10.48	1.74	Neural Network (Specific Model)	11	0.919	3218	322.7874125874126	752315	0.41	715	228	6603	None
+bin_semibin_SASB_HSMA33MG.1	37.55	7.85	Neural Network (Specific Model)	11	0.909	3945	314.83505866114564	1502934	0.43	1449	378	18770	None
+bin_semibin_SASB_HSMA33MG.2	9.86	0.19	Neural Network (Specific Model)	11	0.84	5391	209.15863141524105	478276	0.48	643	94	18806	None
+bin_semibin_SASB_MSM5LLDE.0	6.09	0.04	Neural Network (Specific Model)	11	0.889	3492	214.14640198511165	290530	0.54	403	76	32205	None
+bin_semibin_SASB_MSM5LLDE.1	29.89	0.02	Neural Network (Specific Model)	11	0.914	3803	319.14351851851853	1129847	0.42	1080	302	10395	None
+bin_semibin_SASB_MSM5LLDE.10	6.39	0.04	Neural Network (Specific Model)	11	0.838	6460	243.23076923076923	214350	0.48	247	38	17791	None
--- a/tooldistillator.xml	Thu Oct 17 08:00:15 2024 +0000
+++ b/tooldistillator.xml	Thu Apr 10 15:02:10 2025 +0000
@@ -118,6 +118,19 @@
     --threshold '$tool.select_tool.threshold'
     --read_len '$tool.select_tool.read_len'
     --level '$tool.select_tool.level'
+        #elif $tool.select_tool.tool_list == 'checkm2':
+            #if $tool.select_tool.diamond_results_path
+    --diamond_results_path '$tool.select_tool.diamond_results_path'
+    --diamond_results_hid '$tool.select_tool.diamond_results_path.hid'
+            #end if
+            #if $tool.select_tool.protein_zip_path
+    --protein_zip_path '$tool.select_tool.protein_zip_path'
+    --protein_zip_hid '$tool.select_tool.protein_zip_path.hid'
+            #end if
+            #if $tool.select_tool.checkm2_log_path
+    --checkm2_log_path '$tool.select_tool.checkm2_log_path'
+    --checkm2_log_hid '$tool.select_tool.checkm2_log_path.hid'
+            #end if
         #elif $tool.select_tool.tool_list == 'fastp':
             #if $tool.select_tool.trimmed_forward_R1_path
     --trimmed_forward_R1_path '$tool.select_tool.trimmed_forward_R1_path'
@@ -251,6 +264,7 @@
                         <option value="bandage">Bandage</option>
                         <option value="bracken">Bracken</option>
                         <option value="bwa">BWA</option>
+                        <option value="checkm2">CheckM2</option>
                         <option value="fastp">Fastp</option>
                         <option value="fastqc">Fastqc</option>
                         <option value="filtlong">Filtlong</option>
@@ -326,6 +340,14 @@
                         <expand macro="analysis_software_version"/>
                         <expand macro="reference_database_version"/>
                     </when>
+                    <when value="checkm2">
+                        <param name="input" type="data" format="txt,tabular" multiple="false" label="CheckM2 quality report file" help="Tabular file from CheckM2 output"/>
+                        <param argument="--diamond_results_path" type="data" format="txt,tabular" optional="true" multiple="false" label="CheckM2 Diamond results file" help="DIAMOND_RESULTS file for CheckM2"/>
+                        <param argument="--protein_zip_path" type="data" format="zip,tar,tar.gz" optional="true" multiple="false" label="Archive file with protein files from CheckM2" help="Protein sequence fasta files in an archive file for CheckM2"/>
+                        <param argument="--checkm2_log_path" type="data" format="txt" multiple="false" optional="true" label="CheckM2 log file" help="Log file from CheckM2"/>
+                        <expand macro="analysis_software_version"/>
+                        <expand macro="reference_database_version"/>
+                    </when>
                     <when value="fastp">
                         <param name="input" type="data" format="json" multiple="false" label="Fastp report file" help="Fastp result in json format"/>
                         <param argument="--trimmed_forward_R1_path" type="data" format="fastq,fastq.gz" optional="true" multiple="false" label="Fastp trimmed R1" help="trimmed file R1"/>
@@ -597,6 +619,19 @@
                 </repeat>
                 <repeat name="tools">
                     <conditional name="select_tool">
+                        <param name="tool_list" value="checkm2"/>
+                        <param name="input" value="checkm2/quality_report.tsv" ftype="tabular"/>
+                        <param name="diamond_results_path" value="checkm2/DIAMOND_RESULTS.tsv"/>
+                        <param name="checkm2_log_path" value="checkm2/checkm2.log"/>
+                        <conditional name="origin">
+                            <param name="origin" value="true"/>
+                            <param name="analysis_software_version" value="1.0.2"/>
+                        </conditional>
+                        <param name="reference_database_version" value="Uniref"/>
+                    </conditional>
+                </repeat>
+                <repeat name="tools">
+                    <conditional name="select_tool">
                         <param name="tool_list" value="fastp"/>
                         <param name="input" value="fastp/fastp_report.json" ftype="json"/>
                         <param name="trimmed_forward_R1_path" value="fastp/trimmed_R1.fastq.gz"/>
@@ -771,27 +806,28 @@
                     </conditional>
                 </repeat>
             </section>
-            <output_collection name="output_json" type="list" count="20">
+            <output_collection name="output_json" type="list" count="21">
                 <expand macro="element_assert" name="abricate_0_output" text="Enterococcus faecalis V583"/>
                 <expand macro="element_assert" name="amrfinderplus_1_output" text="mercury resistance transcriptional regulator MerR"/>
                 <expand macro="element_assert" name="bakta_2_output" text="hypothetical protein"/>
                 <expand macro="element_assert" name="bandage_3_output" text="3042326"/>
                 <expand macro="element_assert" name="bracken_4_output" text="Staphylococcus_aureus"/>
                 <expand macro="element_assert" name="bwa_5_output" text="bwa_report"/>
-                <expand macro="element_assert" name="fastp_6_output" text="705053822"/>
-                <expand macro="element_assert" name="fastqc_7_output" text="Sequences flagged as poor quality"/>
-                <expand macro="element_assert" name="filtlong_8_output" text="filtlong_report"/>
-                <expand macro="element_assert" name="flye_9_output" text="contig_2"/>
-                <expand macro="element_assert" name="integronfinder2_10_output" text="2353.0"/>
-                <expand macro="element_assert" name="isescan_11_output" text="IS200/IS605"/>
-                <expand macro="element_assert" name="kraken2_12_output" text="Enterococcus_faecalis"/>
-                <expand macro="element_assert" name="plasmidfinder_13_output" text="repUS40_1_repB"/>
-                <expand macro="element_assert" name="polypolish_14_output" text="320811"/>
-                <expand macro="element_assert" name="quast_15_output" text="2944723"/>
-                <expand macro="element_assert" name="recentrifuge_16_output" text="6.4917947578"/>
-                <expand macro="element_assert" name="refseqmasher_17_output" text="GCF_000519925.1"/>
-                <expand macro="element_assert" name="shovill_18_output" text="436224"/>
-                <expand macro="element_assert" name="tabular_file_19_output" text="NP_814691"/>
+                <expand macro="element_assert" name="checkm2_6_output" text="UniRef100_B5R745"/>
+                <expand macro="element_assert" name="fastp_7_output" text="705053822"/>
+                <expand macro="element_assert" name="fastqc_8_output" text="Sequences flagged as poor quality"/>
+                <expand macro="element_assert" name="filtlong_9_output" text="filtlong_report"/>
+                <expand macro="element_assert" name="flye_10_output" text="contig_2"/>
+                <expand macro="element_assert" name="integronfinder2_11_output" text="2353.0"/>
+                <expand macro="element_assert" name="isescan_12_output" text="IS200/IS605"/>
+                <expand macro="element_assert" name="kraken2_13_output" text="Enterococcus_faecalis"/>
+                <expand macro="element_assert" name="plasmidfinder_14_output" text="repUS40_1_repB"/>
+                <expand macro="element_assert" name="polypolish_15_output" text="320811"/>
+                <expand macro="element_assert" name="quast_16_output" text="2944723"/>
+                <expand macro="element_assert" name="recentrifuge_17_output" text="6.4917947578"/>
+                <expand macro="element_assert" name="refseqmasher_18_output" text="GCF_000519925.1"/>
+                <expand macro="element_assert" name="shovill_19_output" text="436224"/>
+                <expand macro="element_assert" name="tabular_file_20_output" text="NP_814691"/>
             </output_collection>
         </test>
     </tests>
@@ -801,9 +837,10 @@
 ToolDistillator is a tool to extract information from output files of specific tools, expose it as JSON files, and aggregate over several tools.
 
 This tool here is extracting information from output files of specific tools and exposing it as JSON files
+
 ** Tool input**
 
-ToolDistillator can use several input type from at least 21 different tools and 1 generic tool for tabular file:
+ToolDistillator can use several input type from at least 22 different tools and 1 generic tool for tabular file:
 
 +----------------+----------------+-------------------+------------------------------------------------------------+
 | Tools          | Version        | Default input file| Optional files                                             |
@@ -820,6 +857,8 @@
 +----------------+----------------+-------------------+------------------------------------------------------------+
 | BWA            | 2.2.1          |  input1.bam       | input2.bam                                                 |
 +----------------+----------------+-------------------+------------------------------------------------------------+
+| CheckM2        |1.0.2           | quality_report.tsv| DIAMOND_RESULTS.tsv, protein_files.zip, checkm2.log        |
++----------------+----------------+-------------------+------------------------------------------------------------+
 | Fastp          | 0.23.2         |  output.json      |                                                            |
 +----------------+----------------+-------------------+------------------------------------------------------------+
 | Fastqc         | 0.12.1         | report.txt        | report.html                                                |
@@ -850,7 +889,7 @@
 +----------------+----------------+-------------------+------------------------------------------------------------+
 | Staramr        | 0.9.1          |  resfinder.tsv    | mlst.tsv, pointfinder.tsv, plasmidfinder.tsv, settings.tsv |
 +----------------+----------------+-------------------+------------------------------------------------------------+
-| tabular_file   |  0             |  output.tsv       | no optional files                                          |
+| tabular_file   |  0             |  output.tsv       |                                                            |
 +----------------+----------------+-------------------+------------------------------------------------------------+
 
 ** Options **