Dockopt (pydock3 script) - Revision history

Khtang: /* Slurm */

2024-07-18T19:08:00Z

Slurm

← Older revision		Revision as of 19:08, 18 July 2024
Line 115:		Line 115:
	==== Slurm ====		==== Slurm ====

	E.g., on the UCSF Shoichet Lab Gimel cluster (on any node other than 'gimel' itself, such as '~~gimel5~~'):		E.g., on the UCSF Shoichet Lab Gimel cluster (on any node other than 'gimel' itself, such as 'gimel2'):

	export SBATCH_EXEC=/usr/bin/sbatch		export SBATCH_EXEC=/usr/bin/sbatch

Khtang: /* Gimel */

2024-07-18T19:07:43Z

Gimel

← Older revision		Revision as of 19:07, 18 July 2024
Line 14:		Line 14:
	=== Gimel ===		=== Gimel ===

	Only nodes other than ''gimel'' itself are supported, e.g., ''~~gimel5~~''.		Only nodes other than ''gimel'' itself are supported, e.g., ''gimel2''.

	ssh ~~gimel5~~		ssh gimel2
	source /nfs/soft/ian/env.sh		source /nfs/soft/ian/env.sh

Zdingman: Corrected the flag --job_dir_name to --job_dir_path per the dockopt new help page.

2024-06-13T20:33:57Z

Corrected the flag --job_dir_name to --job_dir_path per the dockopt new help page.

← Older revision		Revision as of 20:33, 13 June 2024
Line 46:		Line 46:
	pydock3 dockopt - new		pydock3 dockopt - new

	By default, the job directory is named ''dockopt_job''. To specify a different name, use the "--~~job_dir_name~~" flag. E.g.:		By default, the job directory is named ''dockopt_job''. To specify a different name, use the "--job_dir_path" flag. E.g.:

	pydock3 dockopt - new --~~job_dir_name~~=dockopt_job_2		pydock3 dockopt - new --job_dir_path=dockopt_job_2

	The job directory contains two sub-directories:		The job directory contains two sub-directories:

Ianscottknight: /* Subcommand: run */

2024-05-06T06:53:20Z

Subcommand: run

← Older revision		Revision as of 06:53, 6 May 2024
Line 140:		Line 140:
	Once your job has been configured to your liking, navigate to the the job directory and run ''dockopt'':		Once your job has been configured to your liking, navigate to the the job directory and run ''dockopt'':
	cd <JOB_DIR_NAME>		cd <JOB_DIR_NAME>
	pydock3 dockopt - run <JOB_SCHEDULER_NAME> [--retrodock_job_timeout_minutes=None] [--retrodock_job_max_reattempts=0] [--extra_submission_cmd_params_str=None] [--~~export_negatives_mol2~~=False]		pydock3 dockopt - run <JOB_SCHEDULER_NAME> [--retrodock_job_timeout_minutes=None] [--retrodock_job_max_reattempts=0] [--extra_submission_cmd_params_str=None] [--export_decoys_mol2=False]

	where <JOB_SCHEDULER_NAME> is one of:		where <JOB_SCHEDULER_NAME> is one of:

Ianscottknight at 23:20, 5 May 2024

2024-05-05T23:20:31Z

← Older revision		Revision as of 23:20, 5 May 2024
Line 149:		Line 149:

	Once the dockopt job is complete, the following files will be generated in the job directory:		Once the dockopt job is complete, the following files will be generated in the job directory:
	* ''report.html'': contains (1) a histogram of the performance of all tested docking configurations compared against a distribution of the performance of a random classifier, so as to show whether the test docking configurations are significantly better than ones that can be produced by a random classifier. This is necessary due to the fact that many configurations are being tested. Hence, a Bonferroni correction is applied to the significance threshold, dividing p=0.01 by the number of tested configurations. (2) ROC, charge, and energy plots of the top docking configurations, comparing ~~positive class molecules~~ vs. ~~negative class molecules~~, (3) box plots of enrichment for every multi-valued config parameter, and (4) heatmaps of enrichment for every pair of multi-valued config parameters.		* ''report.html'': contains (1) a histogram of the performance of all tested docking configurations compared against a distribution of the performance of a random classifier, so as to show whether the test docking configurations are significantly better than ones that can be produced by a random classifier. This is necessary due to the fact that many configurations are being tested. Hence, a Bonferroni correction is applied to the significance threshold, dividing p=0.01 by the number of tested configurations. (2) ROC, charge, and energy plots of the top docking configurations, comparing actives vs. decoys, (3) box plots of enrichment for every multi-valued config parameter, and (4) heatmaps of enrichment for every pair of multi-valued config parameters.
	* ''results.csv'': parameter values, criterion values, and other information about each docking configuration.		* ''results.csv'': parameter values, criterion values, and other information about each docking configuration.

Line 157:		Line 157:
	* ''dockfiles/'': parameters files and ''INDOCK'' for given docking configuration		* ''dockfiles/'': parameters files and ''INDOCK'' for given docking configuration
	* ''output/'': contains:		* ''output/'': contains:
	** sub-directories ''~~positives~~/'' (containing ''OUTDOCK'' and ''test.mol2'' files) and ''~~negatives~~/'' (containing just ''OUTDOCK'')		** sub-directories ''actives/'' (containing ''OUTDOCK'' and ''test.mol2'' files) and ''decoys/'' (containing just ''OUTDOCK'')
	* plot images (e.g., ''roc.png'')		* plot images (e.g., ''roc.png'')

	'''Note:''' by default, a mol2 file is exported only for ~~positives~~, not for ~~negatives~~, in order to prevent disk space issues.		'''Note:''' by default, a mol2 file is exported only for actives, not for decoys, in order to prevent disk space issues.

Ianscottknight: /* Subcommand: new */

2024-05-05T23:18:14Z

Subcommand: new

← Older revision		Revision as of 23:18, 5 May 2024
Line 27:		Line 27:
	* ''rec.pdb'' or ''rec.crg.pdb''		* ''rec.pdb'' or ''rec.crg.pdb''
	* ''xtal-lig.pdb''		* ''xtal-lig.pdb''
	* ''~~positives~~.tgz''		* ''actives.tgz''
	* ''~~negatives~~.tgz''		* ''decoys.tgz''

	Note the inclusion of ''~~positives~~.tgz'' and ''~~negatives~~.tgz''. Each of these is a tarball of DB2 files, which may be optionally gzipped (with extension .db2.gz). Each tarball represents a binary class for the binary classification task that the docking model is trained on. The positive class contains the molecules that you want the docking model to preferentially assign ''favorable'' docking scores to. The negative class contains the molecules that you want the docking model to preferentially assign ''unfavorable'' scores to. The most common strategy is to use a set of known ligands as the ~~positive class~~ and a larger set of property-matched decoys for the ~~negative class~~ (a decoy-to-active ratio of 50:1 is standard), but other strategies are supported. For example, to create a docking model that preferentially assigns favorable scores to agonists over antagonists, a set of agonists can be used as the ~~positive class~~ and a set of antagonists can be used for the ~~negative class~~.		Note the inclusion of ''actives.tgz'' and ''actives.tgz''. Each of these is a tarball of DB2 files, which may be optionally gzipped (with extension .db2.gz). Each tarball represents a binary class for the binary classification task that the docking model is trained on. The positive class (actives) contains the molecules that you want the docking model to preferentially assign ''favorable'' docking scores to. The negative class (decoys) contains the molecules that you want the docking model to preferentially assign ''unfavorable'' scores to. The most common strategy is to use a set of known ligands as the actives and a larger set of property-matched decoys for the decoys (a decoy-to-active ratio of 50:1 is standard), but other strategies are supported. For example, to create a docking model that preferentially assigns favorable scores to agonists over antagonists, a set of agonists can be used as the actives and a set of antagonists can be used for the decoys.

	Therefore, you need to build the molecules yourself (see: https://tldr.docking.org/start/build3d38). Each tarball should contain only DB2 files. For example, if one has a directory ''actives/'' containing only DB2 files to use as the ~~positive class~~, then ''~~positives~~.tgz'' should be created as follows:		Therefore, you need to build the molecules yourself (see: https://tldr.docking.org/start/build3d38). Each tarball should contain only DB2 files. For example, if one has a directory ''actives/'' containing only DB2 files to use as the actives, then ''acitves.tgz'' should be created as follows:

	cd actives/		cd actives/
	tar -czf ~~positives~~.tgz .db2		tar -czf actives.tgz .db2

	Similarly, for a directory ''decoys/'' containing only DB2 files to use as ~~the negative class~~, ''~~negatives~~.tgz'' should be created as follows:		Similarly, for a directory ''decoys/'' containing only DB2 files to use as decoys, ''decoys.tgz'' should be created as follows:

	cd decoys/		cd decoys/
	tar -czf ~~negatives~~.tgz .db2		tar -czf decoys.tgz .db2

	To create the file structure for your dockopt job, simply type		To create the file structure for your dockopt job, simply type

Ianscottknight at 23:11, 5 May 2024

2024-05-05T23:11:22Z

← Older revision		Revision as of 23:11, 5 May 2024
Line 19:		Line 19:
	source /nfs/soft/ian/env.sh		source /nfs/soft/ian/env.sh

	== ''new'' ==		== Subcommand: ''new'' ==

	Prepare rec.pdb, xtal-lig.pdb as described in Bender, 2021. https://pubmed.ncbi.nlm.nih.gov/34561691/		Prepare rec.pdb, xtal-lig.pdb as described in Bender, 2021. https://pubmed.ncbi.nlm.nih.gov/34561691/
Line 136:		Line 136:
	export SGE_SETTINGS=/opt/sge/wynton/common/settings.sh		export SGE_SETTINGS=/opt/sge/wynton/common/settings.sh

	== ''run'' ==		== Subcommand: ''run'' ==

	Once your job has been configured to your liking, navigate to the the job directory and run ''dockopt'':		Once your job has been configured to your liking, navigate to the the job directory and run ''dockopt'':

Ianscottknight at 19:54, 1 February 2024

2024-02-01T19:54:56Z

@@ Line 3: / Line 3: @@
 To use DOCK 3.8, you must first license it and install it.
 [[DOCK 3.8:How to install pydock3]]
 == ''new'' ==
@@ Line 146: / Line 161: @@
 '''Note:''' by default, a mol2 file is exported only for positives, not for negatives, in order to prevent disk space issues.

Ianscottknight at 04:35, 2 September 2023

2023-09-02T04:35:59Z

← Older revision		Revision as of 04:35, 2 September 2023
Line 29:		Line 29:
	To create the file structure for your dockopt job, simply type		To create the file structure for your dockopt job, simply type

	pydock3 dockopt - new ~~[--job_dir_name=dockopt_job]~~		pydock3 dockopt - new

	By default, the job directory is named ''dockopt_job''. To specify a different name, use the "--job_dir_name" flag. E.g.:		By default, the job directory is named ''dockopt_job''. To specify a different name, use the "--job_dir_name" flag. E.g.:
Line 139:		Line 139:
	In addition, some number of the best retrodock jobs will be copied to their own sub-directory ''best_retrodock_jobs/''.		In addition, some number of the best retrodock jobs will be copied to their own sub-directory ''best_retrodock_jobs/''.

	Within ''best_retrodock_jobs/'', there are ~~the following files and sub-directories~~:		Within each sub-directory of ''best_retrodock_jobs/'', there are:
	* ''dockfiles/'': parameters files and ''INDOCK'' for given docking configuration		* ''dockfiles/'': parameters files and ''INDOCK'' for given docking configuration
	* ''output/'': contains:		* ''output/'': contains:
	** joblist
	** sub-directories ''positives/'' (containing ''OUTDOCK'' and ''test.mol2'' files) and ''negatives/'' (containing just ''OUTDOCK'')		** sub-directories ''positives/'' (containing ''OUTDOCK'' and ''test.mol2'' files) and ''negatives/'' (containing just ''OUTDOCK'')
	** log files for the retrodock jobs
	* plot images (e.g., ''roc.png'')		* plot images (e.g., ''roc.png'')

Ianscottknight at 22:31, 27 July 2023

2023-07-27T22:31:21Z

← Older revision		Revision as of 22:31, 27 July 2023
Line 15:		Line 15:
	* ''negatives.tgz''		* ''negatives.tgz''

	Note the inclusion of ''positives.tgz'' and ''negatives.tgz''. Each of these is a tarball of DB2 files, which may be ~~optional~~ gzipped (with extension .db2.gz). Each tarball represents a binary class for the binary classification task that the docking model is trained on. The positive class contains the molecules that you want the docking model to preferentially assign ''favorable'' docking scores to. The negative class contains the molecules that you want the docking model to preferentially assign ''unfavorable'' scores to. The most common strategy is to use a set of known ligands as the positive class and a larger set of property-matched decoys for the negative class (a decoy-to-active ratio of 50:1 is standard), but other strategies are supported. For example, to create a docking model that preferentially assigns favorable scores to agonists over antagonists, a set of agonists can be used as the positive class and a set of antagonists can be used for the negative class.		Note the inclusion of ''positives.tgz'' and ''negatives.tgz''. Each of these is a tarball of DB2 files, which may be optionally gzipped (with extension .db2.gz). Each tarball represents a binary class for the binary classification task that the docking model is trained on. The positive class contains the molecules that you want the docking model to preferentially assign ''favorable'' docking scores to. The negative class contains the molecules that you want the docking model to preferentially assign ''unfavorable'' scores to. The most common strategy is to use a set of known ligands as the positive class and a larger set of property-matched decoys for the negative class (a decoy-to-active ratio of 50:1 is standard), but other strategies are supported. For example, to create a docking model that preferentially assigns favorable scores to agonists over antagonists, a set of agonists can be used as the positive class and a set of antagonists can be used for the negative class.

	Therefore, you need to build the molecules yourself (see: https://tldr.docking.org/start/build3d38). Each tarball should contain only DB2 files. For example, if one has a directory ''actives/'' containing only DB2 files to use as the positive class, then ''positives.tgz'' should be created as follows:		Therefore, you need to build the molecules yourself (see: https://tldr.docking.org/start/build3d38). Each tarball should contain only DB2 files. For example, if one has a directory ''actives/'' containing only DB2 files to use as the positive class, then ''positives.tgz'' should be created as follows: