# HG changeset patch
# User bgruening
# Date 1377517051 14400
# Node ID e1bc8415f875c1c791740bc65d4e9d87dc4d9299
Uploaded
diff -r 000000000000 -r e1bc8415f875 chemicaltoolbox_merging_chemical_databases_workflow.ga
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/chemicaltoolbox_merging_chemical_databases_workflow.ga Mon Aug 26 07:37:31 2013 -0400
@@ -0,0 +1,1190 @@
+{
+ "a_galaxy_workflow": "true",
+ "annotation": "Runs the chemicaltoolbox to merge chemical libraries.
\n
\nThis workflow is available on the Galaxy Tool Shed with a README file giving more information, and full citation details (Gruening et al. 2013).",
+ "format-version": "0.1",
+ "name": "Preparation of a large compound library by merging of chemical databases.",
+ "steps": {
+ "0": {
+ "annotation": "",
+ "id": 0,
+ "input_connections": {},
+ "inputs": [],
+ "name": "Online data",
+ "outputs": [
+ {
+ "name": "output",
+ "type": "txt"
+ }
+ ],
+ "position": {
+ "left": 291,
+ "top": 391
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutput": {
+ "action_arguments": {
+ "newtype": "sdf"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "output"
+ },
+ "HideDatasetActionoutput": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "output"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_online_data_fetch",
+ "tool_state": "{\"__page__\": 0, \"whitelist\": \"\\\"\\\"\", \"url_paste\": \"\\\"http://www.drugbank.ca/system/downloads/current/structures/all.sdf.zip\\\"\"}",
+ "tool_version": "0.2",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "1": {
+ "annotation": "",
+ "id": 1,
+ "input_connections": {},
+ "inputs": [],
+ "name": "Online data",
+ "outputs": [
+ {
+ "name": "output",
+ "type": "txt"
+ }
+ ],
+ "position": {
+ "left": 304,
+ "top": 733
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutput": {
+ "action_arguments": {
+ "newtype": "tabular"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "output"
+ },
+ "HideDatasetActionoutput": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "output"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_online_data_fetch",
+ "tool_state": "{\"__page__\": 0, \"whitelist\": \"\\\"\\\"\", \"url_paste\": \"\\\"http://zinc.docking.org/db/bysubset/6/6_p0.smi.gz\\\\nhttp://zinc.docking.org/db/bysubset/46/46_p0.smi.gz\\\"\"}",
+ "tool_version": "0.2",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "2": {
+ "annotation": "",
+ "id": 2,
+ "input_connections": {},
+ "inputs": [],
+ "name": "Online data",
+ "outputs": [
+ {
+ "name": "output",
+ "type": "txt"
+ }
+ ],
+ "position": {
+ "left": 305,
+ "top": 1048
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutput": {
+ "action_arguments": {
+ "newtype": "sdf"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "output"
+ },
+ "HideDatasetActionoutput": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "output"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_online_data_fetch",
+ "tool_state": "{\"__page__\": 0, \"whitelist\": \"\\\"\\\"\", \"url_paste\": \"\\\"ftp://ftp.ebi.ac.uk/pub/databases/chembl/ChEMBLdb/latest/chembl_16.sdf.gz\\\"\"}",
+ "tool_version": "0.2",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "3": {
+ "annotation": "",
+ "id": 3,
+ "input_connections": {},
+ "inputs": [],
+ "name": "PubChem Download",
+ "outputs": [
+ {
+ "name": "pubchem_smi",
+ "type": "smi"
+ }
+ ],
+ "position": {
+ "left": 312,
+ "top": 1448
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionpubchem_smi": {
+ "action_arguments": {
+ "newtype": "tabular"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "pubchem_smi"
+ },
+ "HideDatasetActionpubchem_smi": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "pubchem_smi"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_pubchem_download_as_smiles",
+ "tool_state": "{\"__page__\": 0, \"infile\": \"null\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "4": {
+ "annotation": "",
+ "id": 4,
+ "input_connections": {
+ "infile": {
+ "id": 0,
+ "output_name": "output"
+ }
+ },
+ "inputs": [],
+ "name": "Change Title",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "sdf"
+ }
+ ],
+ "position": {
+ "left": 584,
+ "top": 297
+ },
+ "post_job_actions": {
+ "HideDatasetActionoutfile": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_change_title",
+ "tool_state": "{\"__page__\": 0, \"infile\": \"null\", \"key\": \"\\\"DRUGBANK_ID\\\"\"}",
+ "tool_version": "0.0.1",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "5": {
+ "annotation": "",
+ "id": 5,
+ "input_connections": {
+ "input": {
+ "id": 1,
+ "output_name": "output"
+ }
+ },
+ "inputs": [],
+ "name": "Filter",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 593,
+ "top": 624
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionout_file1": {
+ "action_arguments": {
+ "newtype": "smi"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "out_file1"
+ },
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "Filter1",
+ "tool_state": "{\"input\": \"null\", \"header_lines\": \"\\\"0\\\"\", \"cond\": \"\\\"c1.strip()!='' and c1.strip()!='.'\\\"\", \"__page__\": 0}",
+ "tool_version": "1.1.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "6": {
+ "annotation": "",
+ "id": 6,
+ "input_connections": {
+ "infile": {
+ "id": 2,
+ "output_name": "output"
+ }
+ },
+ "inputs": [],
+ "name": "Change Title",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "sdf"
+ }
+ ],
+ "position": {
+ "left": 592,
+ "top": 988
+ },
+ "post_job_actions": {
+ "HideDatasetActionoutfile": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_change_title",
+ "tool_state": "{\"__page__\": 0, \"infile\": \"null\", \"key\": \"\\\"chembl_id\\\"\"}",
+ "tool_version": "0.0.1",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "7": {
+ "annotation": "",
+ "id": 7,
+ "input_connections": {
+ "input": {
+ "id": 3,
+ "output_name": "pubchem_smi"
+ }
+ },
+ "inputs": [],
+ "name": "Filter",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 623,
+ "top": 1365
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionout_file1": {
+ "action_arguments": {
+ "newtype": "smi"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "out_file1"
+ },
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "Filter1",
+ "tool_state": "{\"input\": \"null\", \"header_lines\": \"\\\"0\\\"\", \"cond\": \"\\\"c1.strip()!='' and c1.strip()!='.'\\\"\", \"__page__\": 0}",
+ "tool_version": "1.1.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "8": {
+ "annotation": "",
+ "id": 8,
+ "input_connections": {
+ "infile": {
+ "id": 4,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Compound Convert",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "text"
+ }
+ ],
+ "position": {
+ "left": 537,
+ "top": 410
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "tabular"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "HideDatasetActionoutfile": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_compound_convert",
+ "tool_state": "{\"__page__\": 0, \"dative_bonds\": \"\\\"True\\\"\", \"appendtotitle\": \"\\\"\\\"\", \"remove_h\": \"\\\"False\\\"\", \"oformat\": \"{\\\"can_iso_chi\\\": \\\"False\\\", \\\"can_atomclass_out\\\": \\\"False\\\", \\\"can_rad\\\": \\\"False\\\", \\\"can_exp_h\\\": \\\"False\\\", \\\"oformat_opts_selector\\\": \\\"can\\\", \\\"__current_case__\\\": 10}\", \"ph\": \"\\\"7.0\\\"\", \"unique\": \"{\\\"unique_opts_selector\\\": \\\"\\\", \\\"__current_case__\\\": 0}\", \"infile\": \"null\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "9": {
+ "annotation": "",
+ "id": 9,
+ "input_connections": {
+ "infile": {
+ "id": 5,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Compound Convert",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "text"
+ }
+ ],
+ "position": {
+ "left": 537,
+ "top": 741
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "tabular"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "HideDatasetActionoutfile": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_compound_convert",
+ "tool_state": "{\"__page__\": 0, \"dative_bonds\": \"\\\"True\\\"\", \"appendtotitle\": \"\\\"\\\"\", \"remove_h\": \"\\\"False\\\"\", \"oformat\": \"{\\\"can_iso_chi\\\": \\\"False\\\", \\\"can_atomclass_out\\\": \\\"False\\\", \\\"can_rad\\\": \\\"False\\\", \\\"can_exp_h\\\": \\\"False\\\", \\\"oformat_opts_selector\\\": \\\"can\\\", \\\"__current_case__\\\": 10}\", \"ph\": \"\\\"7.0\\\"\", \"unique\": \"{\\\"unique_opts_selector\\\": \\\"\\\", \\\"__current_case__\\\": 0}\", \"infile\": \"null\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "10": {
+ "annotation": "",
+ "id": 10,
+ "input_connections": {
+ "infile": {
+ "id": 6,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Compound Convert",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "text"
+ }
+ ],
+ "position": {
+ "left": 557,
+ "top": 1115
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "tabular"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "HideDatasetActionoutfile": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_compound_convert",
+ "tool_state": "{\"__page__\": 0, \"dative_bonds\": \"\\\"True\\\"\", \"appendtotitle\": \"\\\"\\\"\", \"remove_h\": \"\\\"False\\\"\", \"oformat\": \"{\\\"can_iso_chi\\\": \\\"False\\\", \\\"can_atomclass_out\\\": \\\"False\\\", \\\"can_rad\\\": \\\"False\\\", \\\"can_exp_h\\\": \\\"False\\\", \\\"oformat_opts_selector\\\": \\\"can\\\", \\\"__current_case__\\\": 10}\", \"ph\": \"\\\"7.0\\\"\", \"unique\": \"{\\\"unique_opts_selector\\\": \\\"\\\", \\\"__current_case__\\\": 0}\", \"infile\": \"null\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "11": {
+ "annotation": "",
+ "id": 11,
+ "input_connections": {
+ "infile": {
+ "id": 7,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Compound Convert",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "text"
+ }
+ ],
+ "position": {
+ "left": 576,
+ "top": 1480
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "tabular"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "HideDatasetActionoutfile": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_compound_convert",
+ "tool_state": "{\"__page__\": 0, \"dative_bonds\": \"\\\"True\\\"\", \"appendtotitle\": \"\\\"\\\"\", \"remove_h\": \"\\\"False\\\"\", \"oformat\": \"{\\\"can_iso_chi\\\": \\\"False\\\", \\\"can_atomclass_out\\\": \\\"False\\\", \\\"can_rad\\\": \\\"False\\\", \\\"can_exp_h\\\": \\\"False\\\", \\\"oformat_opts_selector\\\": \\\"can\\\", \\\"__current_case__\\\": 10}\", \"ph\": \"\\\"7.0\\\"\", \"unique\": \"{\\\"unique_opts_selector\\\": \\\"\\\", \\\"__current_case__\\\": 0}\", \"infile\": \"null\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "12": {
+ "annotation": "",
+ "id": 12,
+ "input_connections": {
+ "input": {
+ "id": 8,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Add column",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 951,
+ "top": 295
+ },
+ "post_job_actions": {
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "addValue",
+ "tool_state": "{\"__page__\": 0, \"input\": \"null\", \"exp\": \"\\\"_drugbank\\\"\", \"iterate\": \"\\\"no\\\"\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "13": {
+ "annotation": "",
+ "id": 13,
+ "input_connections": {
+ "input": {
+ "id": 9,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Add column",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 963,
+ "top": 635
+ },
+ "post_job_actions": {
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "addValue",
+ "tool_state": "{\"__page__\": 0, \"input\": \"null\", \"exp\": \"\\\"_zinc\\\"\", \"iterate\": \"\\\"no\\\"\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "14": {
+ "annotation": "",
+ "id": 14,
+ "input_connections": {
+ "input": {
+ "id": 10,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Add column",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 981,
+ "top": 987
+ },
+ "post_job_actions": {
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "addValue",
+ "tool_state": "{\"__page__\": 0, \"input\": \"null\", \"exp\": \"\\\"_chembldb\\\"\", \"iterate\": \"\\\"no\\\"\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "15": {
+ "annotation": "",
+ "id": 15,
+ "input_connections": {
+ "input": {
+ "id": 11,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Add column",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 995,
+ "top": 1358
+ },
+ "post_job_actions": {
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "addValue",
+ "tool_state": "{\"__page__\": 0, \"input\": \"null\", \"exp\": \"\\\"_pubchem\\\"\", \"iterate\": \"\\\"no\\\"\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "16": {
+ "annotation": "",
+ "id": 16,
+ "input_connections": {
+ "input": {
+ "id": 12,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Unique",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 960,
+ "top": 420
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "smi"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "RenameDatasetActionoutfile": {
+ "action_arguments": {
+ "newname": "Non-redundant DrugBank"
+ },
+ "action_type": "RenameDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "bg_uniq",
+ "tool_state": "{\"adv_opts\": \"{\\\"column_end\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"column_start\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"__current_case__\\\": 1}\", \"input\": \"null\", \"ignore_case\": \"\\\"False\\\"\", \"is_numeric\": \"\\\"False\\\"\", \"__page__\": 0}",
+ "tool_version": "0.3",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "17": {
+ "annotation": "",
+ "id": 17,
+ "input_connections": {
+ "input": {
+ "id": 13,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Unique",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 955,
+ "top": 757
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "smi"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "RenameDatasetActionoutfile": {
+ "action_arguments": {
+ "newname": "Non-redundant ZINC"
+ },
+ "action_type": "RenameDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "bg_uniq",
+ "tool_state": "{\"adv_opts\": \"{\\\"column_end\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"column_start\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"__current_case__\\\": 1}\", \"input\": \"null\", \"ignore_case\": \"\\\"False\\\"\", \"is_numeric\": \"\\\"False\\\"\", \"__page__\": 0}",
+ "tool_version": "0.3",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "18": {
+ "annotation": "",
+ "id": 18,
+ "input_connections": {
+ "input": {
+ "id": 14,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Unique",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 981,
+ "top": 1122
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "smi"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "RenameDatasetActionoutfile": {
+ "action_arguments": {
+ "newname": "Non-redundant ChEMBLdb"
+ },
+ "action_type": "RenameDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "bg_uniq",
+ "tool_state": "{\"adv_opts\": \"{\\\"column_end\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"column_start\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"__current_case__\\\": 1}\", \"input\": \"null\", \"ignore_case\": \"\\\"False\\\"\", \"is_numeric\": \"\\\"False\\\"\", \"__page__\": 0}",
+ "tool_version": "0.3",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "19": {
+ "annotation": "",
+ "id": 19,
+ "input_connections": {
+ "input1": {
+ "id": 12,
+ "output_name": "out_file1"
+ },
+ "queries_0|input2": {
+ "id": 13,
+ "output_name": "out_file1"
+ },
+ "queries_1|input2": {
+ "id": 14,
+ "output_name": "out_file1"
+ },
+ "queries_2|input2": {
+ "id": 15,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Concatenate datasets",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 1419,
+ "top": 824
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionout_file1": {
+ "action_arguments": {
+ "newtype": "tabular"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "out_file1"
+ },
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "cat1",
+ "tool_state": "{\"__page__\": 0, \"input1\": \"null\", \"queries\": \"[{\\\"input2\\\": null, \\\"__index__\\\": 0}, {\\\"input2\\\": null, \\\"__index__\\\": 1}, {\\\"input2\\\": null, \\\"__index__\\\": 2}]\"}",
+ "tool_version": "1.0.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "20": {
+ "annotation": "",
+ "id": 20,
+ "input_connections": {
+ "input": {
+ "id": 15,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Unique",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 1001,
+ "top": 1476
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "smi"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "RenameDatasetActionoutfile": {
+ "action_arguments": {
+ "newname": "Non-redundant PubChem"
+ },
+ "action_type": "RenameDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "bg_uniq",
+ "tool_state": "{\"adv_opts\": \"{\\\"column_end\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"column_start\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"__current_case__\\\": 1}\", \"input\": \"null\", \"ignore_case\": \"\\\"False\\\"\", \"is_numeric\": \"\\\"False\\\"\", \"__page__\": 0}",
+ "tool_version": "0.3",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "21": {
+ "annotation": "",
+ "id": 21,
+ "input_connections": {
+ "input1": {
+ "id": 19,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Merge Columns",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "tabular"
+ }
+ ],
+ "position": {
+ "left": 1763,
+ "top": 907
+ },
+ "post_job_actions": {
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "mergeCols1",
+ "tool_state": "{\"__page__\": 0, \"input1\": \"null\", \"col2\": \"{\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"3\\\"}\", \"columns\": \"[]\", \"col1\": \"{\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"2\\\"}\"}",
+ "tool_version": "1.0.1",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "22": {
+ "annotation": "",
+ "id": 22,
+ "input_connections": {
+ "input": {
+ "id": 21,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Cut",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "tabular"
+ }
+ ],
+ "position": {
+ "left": 1763,
+ "top": 797
+ },
+ "post_job_actions": {
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "Cut1",
+ "tool_state": "{\"columnList\": \"\\\"c1,c4\\\"\", \"input\": \"null\", \"delimiter\": \"\\\"T\\\"\", \"__page__\": 0}",
+ "tool_version": "1.0.2",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "23": {
+ "annotation": "",
+ "id": 23,
+ "input_connections": {
+ "input": {
+ "id": 22,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Filter",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 2094,
+ "top": 912
+ },
+ "post_job_actions": {
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "Filter1",
+ "tool_state": "{\"input\": \"null\", \"header_lines\": \"\\\"0\\\"\", \"cond\": \"\\\"c1.strip()!='' and c1.strip()!='.'\\\"\", \"__page__\": 0}",
+ "tool_version": "1.1.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "24": {
+ "annotation": "",
+ "id": 24,
+ "input_connections": {
+ "input": {
+ "id": 23,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Manipulation",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 2099,
+ "top": 804
+ },
+ "post_job_actions": {
+ "HideDatasetActionoutfile": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "sed_stream_editor",
+ "tool_state": "{\"__page__\": 0, \"input\": \"null\", \"pattern\": \"\\\"s/ /_/g\\\"\"}",
+ "tool_version": "0.0.1",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "25": {
+ "annotation": "",
+ "id": 25,
+ "input_connections": {
+ "input": {
+ "id": 24,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Unique",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 2098,
+ "top": 681
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "smi"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "RenameDatasetActionoutfile": {
+ "action_arguments": {
+ "newname": "Non-redundant merged libraries"
+ },
+ "action_type": "RenameDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "bg_uniq",
+ "tool_state": "{\"adv_opts\": \"{\\\"column_end\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"column_start\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"__current_case__\\\": 1}\", \"input\": \"null\", \"ignore_case\": \"\\\"False\\\"\", \"is_numeric\": \"\\\"False\\\"\", \"__page__\": 0}",
+ "tool_version": "0.3",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "26": {
+ "annotation": "",
+ "id": 26,
+ "input_connections": {
+ "infile": {
+ "id": 25,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Remove counterions and fragments",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "sdf"
+ }
+ ],
+ "position": {
+ "left": 2392,
+ "top": 907
+ },
+ "post_job_actions": {
+ "HideDatasetActionoutfile": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "ctb_remIons",
+ "tool_state": "{\"__page__\": 0, \"infile\": \"null\"}",
+ "tool_version": "1.0",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "27": {
+ "annotation": "",
+ "id": 27,
+ "input_connections": {
+ "input": {
+ "id": 26,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Select",
+ "outputs": [
+ {
+ "name": "out_file1",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 2447,
+ "top": 795
+ },
+ "post_job_actions": {
+ "HideDatasetActionout_file1": {
+ "action_arguments": {},
+ "action_type": "HideDatasetAction",
+ "output_name": "out_file1"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "Grep1",
+ "tool_state": "{\"__page__\": 0, \"input\": \"null\", \"invert\": \"\\\"true\\\"\", \"pattern\": \"\\\"\\\\\\\\*\\\"\"}",
+ "tool_version": "1.0.1",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "28": {
+ "annotation": "",
+ "id": 28,
+ "input_connections": {
+ "input": {
+ "id": 27,
+ "output_name": "out_file1"
+ }
+ },
+ "inputs": [],
+ "name": "Unique",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 2444,
+ "top": 678
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "tabular"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "RenameDatasetActionoutfile": {
+ "action_arguments": {
+ "newname": "Filtered merged libraries"
+ },
+ "action_type": "RenameDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "bg_uniq",
+ "tool_state": "{\"adv_opts\": \"{\\\"adv_opts_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"input\": \"null\", \"ignore_case\": \"\\\"False\\\"\", \"is_numeric\": \"\\\"False\\\"\", \"__page__\": 0}",
+ "tool_version": "0.3",
+ "type": "tool",
+ "user_outputs": []
+ },
+ "29": {
+ "annotation": "",
+ "id": 29,
+ "input_connections": {
+ "input": {
+ "id": 28,
+ "output_name": "outfile"
+ }
+ },
+ "inputs": [],
+ "name": "Unique",
+ "outputs": [
+ {
+ "name": "outfile",
+ "type": "input"
+ }
+ ],
+ "position": {
+ "left": 2444,
+ "top": 562
+ },
+ "post_job_actions": {
+ "ChangeDatatypeActionoutfile": {
+ "action_arguments": {
+ "newtype": "smi"
+ },
+ "action_type": "ChangeDatatypeAction",
+ "output_name": "outfile"
+ },
+ "RenameDatasetActionoutfile": {
+ "action_arguments": {
+ "newname": "Non-redundant filtered merged libraries"
+ },
+ "action_type": "RenameDatasetAction",
+ "output_name": "outfile"
+ }
+ },
+ "tool_errors": null,
+ "tool_id": "bg_uniq",
+ "tool_state": "{\"adv_opts\": \"{\\\"column_end\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"column_start\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"1\\\"}, \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"__current_case__\\\": 1}\", \"input\": \"null\", \"ignore_case\": \"\\\"False\\\"\", \"is_numeric\": \"\\\"False\\\"\", \"__page__\": 0}",
+ "tool_version": "0.3",
+ "type": "tool",
+ "user_outputs": []
+ }
+ }
+}
diff -r 000000000000 -r e1bc8415f875 readme.rst
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/readme.rst Mon Aug 26 07:37:31 2013 -0400
@@ -0,0 +1,44 @@
+This package is a Galaxy workflow for merging of chemical libraries using the chemicaltoolbox.
+
+To learn more about the chemicaltoolbox and all the integrated tools,
+please have a look at https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox.
+
+
+Sample Data
+===========
+
+The workflow will fetch the sample data from online resources, like PubChem, DrugBank.
+
+
+Citation
+========
+
+If you use this workflow directly, or a derivative of it, or the associated
+chemicaltoolbox wrappers for Galaxy, in work leading to a scientific publication,
+please cite:
+
+Grüning, B. et al. (2013)
+ChemicalToolBoX, a new Galaxy for cheminformatics (Submitted).
+
+
+Availability
+============
+
+This workflow is available on the main Galaxy Tool Shed:
+
+ http://toolshed.g2.bx.psu.edu/view/bgruening/chemicaltoolbox_merging_chemical_databases_workflow
+
+Development is being done on github:
+
+https://github.com/bgruening/galaxytools/workflows/chemicaltoolbox
+
+
+Dependencies
+============
+
+These dependencies should be resolved automatically via the Galaxy Tool Shed:
+
+* http://toolshed.g2.bx.psu.edu/view/bgruening/unique
+* http://toolshed.g2.bx.psu.edu/view/bjoern-gruening/sed_wrapper
+* http://toolshed.g2.bx.psu.edu/view/bgruening/chemicaltoolbox
+
diff -r 000000000000 -r e1bc8415f875 repository_dependencies.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/repository_dependencies.xml Mon Aug 26 07:37:31 2013 -0400
@@ -0,0 +1,6 @@
+
+
+
+
+
+