hERG Inhibition - Revision history

Kirilas: /* hERG IC50 */

2023-07-26T09:45:45Z

hERG IC50

← Older revision		Revision as of 09:45, 26 July 2023
Line 40:		Line 40:

	===hERG IC50===		===hERG IC50===
	[[Image:Herg_IC50\|center]]		[[Image:Herg_IC50.png\|center]]
	<br />		<br />

Line 50:		Line 50:
	# A "heatmap" plot illustrating the partial dependences of hERG inhibition potential on lipophilicity and ionization parameters.		# A "heatmap" plot illustrating the partial dependences of hERG inhibition potential on lipophilicity and ionization parameters.

	[[Image:Herg_heatmap\|left]]		[[Image:Herg_heatmap.png\|left]]

	:a. The cyan-colored dot indicates the position of the current compound.		:a. The cyan-colored dot indicates the position of the current compound.

Kirilas: Updated for 2023 release

2023-07-26T09:42:16Z

Updated for 2023 release

← Older revision		Revision as of 09:42, 26 July 2023
Line 9:		Line 9:
	===Features===		===Features===
	* Predicts the probability for a compound to inhibit hERG channel at clinically relevant concentrations (K<sub>i</sub> < 10 μM).		* Predicts the probability for a compound to inhibit hERG channel at clinically relevant concentrations (K<sub>i</sub> < 10 μM).
	* Predictions are based on a data set of ~~more than 6600~~ compounds with experimental results collected from published hERG inhibition studies utilizing either patch-clamp or competitive binding methods.		* Predictions are based on a data set of almost 9400 compounds with experimental results collected from published hERG inhibition studies utilizing either patch-clamp or competitive binding methods.
	* Calculates Reliability Index (RI values) of predictions that indicates whether tested compounds belong to Applicability Domain of predictive model.		* Calculates Reliability Index (RI values) of predictions that indicates whether tested compounds belong to Applicability Domain of predictive model.
	* Performs a similarity search and displays top 5 most similar structures from the training set of the model along with their names, experimental results, and literature references.		* Performs a similarity search and displays top 5 most similar structures from the training set of the model along with their names, experimental results, and literature references.
	* Supports training the model using ‘in-house’ data, including those generated by ‘in-house’ screening protocol.		* Supports training the model using ‘in-house’ data, including those generated by ‘in-house’ screening protocol.
	* ~~Additional PhysChem model~~ enables ~~the user to explore~~ the influence of key physicochemical properties of drugs on their hERG ~~inhibition potential~~.		* [[hERG_Inhibition#hERG_IC50\|hERG_IC50]] module enables exploration of the influence of key physicochemical properties of drugs on their hERG liability and provides a quantitative estimate of inhibitory potency in the form of predicted IC50 value.

	~~<!--~~<span style="color:red; font-weight: bold;">IMPORTANT NOTE:</span>		<span style="color:red; font-weight: bold;">IMPORTANT NOTE:</span>

	If you installed ~~ACD/~~Percepta ~~v. 2016~~ as an in-place upgrade ~~over the previous version~~, hERG Inhibition module will continue using the same set of Self-training libraries that was configured in your previous installation. ~~To take advantage of~~ the new, significantly extended built-in library that comes with the ~~current version~~ of ~~the software~~, you will need to click "Configure" and manually select the following entry: ''hERG-I (Ki less than 10 uM) v. 1.3 (read-only)''.		If you installed Percepta as an in-place upgrade with Expert user privileges, hERG Inhibition module will continue using the same set of Self-training libraries that was configured in your previous installation. If you are upgrading over version 2021 or earlier, this will not include the new, significantly extended built-in library that comes with the software since v. 2022. To take advantage of this library, you will need to click "Configure" and manually select the following entry: ''hERG-I (Ki less than 10 uM) v. 1.4 (read-only)''.

	In case of a clean installation, this library is selected automatically, and no further action is required.~~-->~~		In case of a clean installation, or an upgrade with a Limited user key, this library is selected automatically, and no further action is required.
	<br />		<br />

Line 25:		Line 25:
	<br />		<br />

			===hERG Inhibitors===
	[[Image:Herg_inhibition.png\|center]]		[[Image:Herg_inhibition.png\|center]]
	<br />		<br />
Line 34:		Line 35:
	#* RI in range 0.5-0.75 – Moderate Reliability,		#* RI in range 0.5-0.75 – Moderate Reliability,
	#* RI >= 0.75 – High Reliability		#* RI >= 0.75 – High Reliability
	# "Configure" and "Train" buttons provide the means to select the training library for use in calculations and to add new data to that library. The ~~name~~ of the currently selected ~~library is~~ indicated with italic font		# "Configure" and "Train" buttons provide the means to select the training library for use in calculations and to add new data to that library. The names of the currently selected libraries are indicated with italic font
	# ~~Switch to the PhysChem model~~		# 5 most similar structures from the self-training library with compound names, experimental classification (Inhibitor or Non-inhibitor), and references
	~~# Up to~~ 5 similar structures in the training ~~set~~ with names, experimental ~~results~~ (Inhibitor, Non-inhibitor~~, Inconclusive data~~), and references
	<br />		<br />

	[[Image:~~Herg_inhibition_PhysChem.png~~\|center]]		===hERG IC50===
			[[Image:Herg_IC50\|center]]
	<br />		<br />

	# Main physicochemical descriptors (LogP, Molecular Weight, Acid and Base pKa) that are used as inputs for calculation. By default, automatically calculated values are used. Enter experimental values to improve prediction accuracy or type custom values to model the influence of the respective properties on hERG inhibitory potential of the compound of interest.		# Main physicochemical descriptors (LogP, Molecular Weight, Acid and Base pKa) that are used as inputs for calculation. By default, automatically calculated values are used. Enter experimental values to improve prediction accuracy or type custom values to model the influence of the respective properties on hERG inhibitory potential of the compound of interest.
	# Click the "Undo" button to revert the manually entered property value to the automatically calculated value (LogP and pKa (Acid) in this picture) for a compound and to recalculate ~~the probability of~~ hERG inhibition		# Click the "Undo" button to revert the manually entered property value to the automatically calculated value (LogP and pKa (Acid) in this picture) for a compound and to recalculate hERG inhibition potency
	# Click to recalculate ~~the probability of~~ hERG inhibition using the currently specified parameter values		# Click to recalculate hERG inhibition potency using the currently specified parameter values
	# ~~Click to select the experimental method to simulate in calculations:~~		# Estimated hERG half-inhibitory constant (IC<sub>50</sub>) of the compound
	~~#* Conventional patch~~-~~clamp~~		# 5 most similar structures from the training set with compound names, experimental results (exact IC50 values or intervals), type of assay used for meaurements, and references
	~~#* Automated patch-clamp~~		# A "heatmap" plot illustrating the partial dependences of hERG inhibition potential on lipophilicity and ionization parameters.
	~~#* Dofetilide replacement~~
	~~#* Astemizole replacement~~		[[Image:Herg_heatmap\|left]]
	~~#* MK-400 replacement~~
	~~#* Other~~/~~Not specified~~		:a. The cyan-colored dot indicates the position of the current compound.
	~~# Estimated probability~~ of the compound ~~being a hERG channel inhibitor~~
	# ~~Switch to~~ the ~~GALAS model~~		:b. Click to select the variables to plot:
	# A "heatmap" plot illustrating the partial dependences of hERG inhibition potential on lipophilicity and ionization parameters. The ~~light blue~~ dot indicates the position of the current compound.
	# Click to select the variables to plot:		* LogP vs pKa (Acid)
	#* LogP vs pKa (Acid)		* LogP vs pKa (Base)
	#* LogP vs pKa (Base)		* pKa (Acid) vs pKa (Base)
	#* pKa (Acid) vs pKa (Base)
	<br />		<br style="clear:both">

	<div class="mw-collapsible">		<div class="mw-collapsible">
Line 68:		Line 69:
	<div class="mw-collapsible-content">		<div class="mw-collapsible-content">
	===Experimental data & predictive models===		===Experimental data & predictive models===
	The built-in library included in hERG inhibition module contains ~~6670~~ compounds with experimental values determined using patch-clamp (conventional and automatic) and competitive radioligand displacement assays (reference ligands: dofetilide, astemizole, MK-499). The data were collected from ChEMBL database, as well as original literature publications. Detailed information about the employed data collection and processing procedures can be found in ''J Comput Aided Mol Des.'' '''2016''';30(12):1175-1188. [https://doi.org/10.1007/s10822-016-9986-0]		The current iteration of the built-in library included in hERG inhibition module contains 9383 compounds with experimental values determined using patch-clamp (conventional and automatic) and competitive radioligand displacement assays (reference ligands: dofetilide, astemizole, MK-499). The data were collected from ChEMBL database, as well as original literature publications. Detailed information about the employed data collection and processing procedures can be found in ''J Comput Aided Mol Des.'' '''2016''';30(12):1175-1188. [https://doi.org/10.1007/s10822-016-9986-0]

	~~The~~ module contains two different ~~probabilistic~~ models:		hERG Inhibition module group contains two different models:
	* PhysChem model - a ~~GBM (~~Gradient Boosting ~~Machine~~) ~~classifier~~ that ~~was built~~ using ~~the full~~ data ~~set and~~ a minimal set of physicochemical descriptors including octanol/water LogP, acid and base pKa, molecular size, topology and flexibility parameters~~, as well as an indicator variable denoting the type of assay used to determine a particular value~~. Technical information about the development procedures and performance of this model is available in ''J Comput Aided Mol Des.'' '''~~2016~~''';30(12):~~1175~~-~~1188~~. [https://doi.org/10.1007/s10822-~~016~~-~~9986~~-0]		* hERG IC50 module presents a PhysChem-based quantitative model - a Gradient Boosting AFT (Accelerated Failure Time) model that has been trained using both fully quantitative and censored (interval) data. This model predicts predicts IC<sub>50</sub> values from a minimal set of physicochemical descriptors including octanol/water LogP, acid and base pKa, molecular size, topology and flexibility parameters. Technical information about the development procedures and performance of this model is available in ''J Comput Aided Mol Des.'' '''2022''';36(12):837-849. [https://doi.org/10.1007/s10822-022-00483-0]
	* GALAS model - a two-part trainable approach, involving a 'baseline' statistical fragmental model, and a similarity-based correction routine that forms the basis of trainability. Originally, this model was built using a smaller data set of 663 molecules with high quality quantitative data. The information provided below applies to that model. In the current version of the software, the original 'baseline' model has been trained using the full database of ~~6670~~ molecules to ensure the best possible coverage of pharmaceutically relevant chemical space.		* hERG Inhibitors module provides a probabilistic GALAS model, which is based on a two-part trainable approach, involving a 'baseline' statistical fragmental model, and a similarity-based correction routine that forms the basis of trainability. Originally, this model was built using a smaller data set of 663 molecules with high quality quantitative data. The information provided below applies to that initial model. In the current version of the software, the original 'baseline' model has been trained using the full database of 9383 molecules to ensure the best possible coverage of pharmaceutically relevant chemical space.

	===Description of the original GALAS model===		===Description of the original GALAS model===

Kirilas: Updated for 2020 release

2020-08-18T08:34:33Z

Updated for 2020 release

Show changes

Kirilas: Removed new library warning

2018-10-30T10:40:19Z

Removed new library warning

← Older revision		Revision as of 10:40, 30 October 2018
Line 14:		Line 14:
	* Training of the model using ‘in-house’ data generated by ‘in-house’ screening protocol.		* Training of the model using ‘in-house’ data generated by ‘in-house’ screening protocol.

	<span style="color:red; font-weight: bold;">IMPORTANT NOTE:</span>		<!--<span style="color:red; font-weight: bold;">IMPORTANT NOTE:</span>

	If you installed ACD/Percepta v. 2016 as an in-place upgrade over the previous version, hERG Inhibition module will continue using the same set of Self-training libraries that was configured in your previous installation. To take advantage of the new, significantly extended built-in library that comes with the current version of the software, you will need to click "Configure" and manually select the folowing entry: ''hERG-I (Ki less than 10 uM) v. 1.3 (read-only)''.		If you installed ACD/Percepta v. 2016 as an in-place upgrade over the previous version, hERG Inhibition module will continue using the same set of Self-training libraries that was configured in your previous installation. To take advantage of the new, significantly extended built-in library that comes with the current version of the software, you will need to click "Configure" and manually select the folowing entry: ''hERG-I (Ki less than 10 uM) v. 1.3 (read-only)''.

	In case of a clean installation, this library is selected automatically, and no further action is required.		In case of a clean installation, this library is selected automatically, and no further action is required.-->
	<br />		<br />

Kirilas at 10:00, 15 June 2017

2017-06-15T10:00:50Z

← Older revision		Revision as of 10:00, 15 June 2017
Line 36:		Line 36:
	<br />		<br />

	<div class="mw-collapsible ~~mw-collapsed~~">		<div class="mw-collapsible">

	==Technical information==		==Technical information==

Kirilas: /* Experimental data */

2016-05-25T11:03:32Z

Experimental data

← Older revision		Revision as of 11:03, 25 May 2016
Line 43:		Line 43:
	<div class="mw-collapsible-content">		<div class="mw-collapsible-content">
	===Experimental data===		===Experimental data===
	The built-in library included in hERG inhibition module contains 6670 compounds with experimental values determined using patch-clamp (conventional and automatic) and competitive radioligand displacement assays (reference ligands: (dofetilide, astemizole, MK-499). The data were collected from ChEMBL database, as well as original literature publications.		The built-in library included in hERG inhibition module contains 6670 compounds with experimental values determined using patch-clamp (conventional and automatic) and competitive radioligand displacement assays (reference ligands: dofetilide, astemizole, MK-499). The data were collected from ChEMBL database, as well as original literature publications.

	The information provided below applies to the original 'baseline' model that was parameterized on a smaller data set of 663 molecules with high quality quantitative data. However, predictions performed by the module use this 'baseline' model trained with full database of 6670 molecules to ensure the best possible coverage of pharmaceutically relevant chemical space.		The information provided below applies to the original 'baseline' model that was parameterized on a smaller data set of 663 molecules with high quality quantitative data. However, predictions performed by the module use this 'baseline' model trained with full database of 6670 molecules to ensure the best possible coverage of pharmaceutically relevant chemical space.

Kirilas: /* Experimental data */

2016-05-25T11:02:58Z

Experimental data

← Older revision		Revision as of 11:02, 25 May 2016
Line 43:		Line 43:
	<div class="mw-collapsible-content">		<div class="mw-collapsible-content">
	===Experimental data===		===Experimental data===
	~~Data set used for model development consisted of 663 binary~~ values ~~(inhibitor, non-inhibitor) These were collected from original publications considering two types of experiments:~~		The built-in library included in hERG inhibition module contains 6670 compounds with experimental values determined using patch-clamp (conventional and automatic) and competitive radioligand displacement assays (reference ligands: (dofetilide, astemizole, MK-499). The data were collected from ChEMBL database, as well as original literature publications.
	* Electrophysiological patch-clamp ~~assay - hERG current inhibition expressed as IC50 constants~~ (~~512 compounds~~).
	* Radioligand (dofetilide, astemizole, MK-499) ~~displacement assay providing~~ ''~~K<sub>i</sub>~~'' ~~values (161 compound)~~.		The information provided below applies to the original 'baseline' model that was parameterized on a smaller data set of 663 molecules with high quality quantitative data. However, predictions performed by the module use this 'baseline' model trained with full database of 6670 molecules to ensure the best possible coverage of pharmaceutically relevant chemical space.

	===Assignment of qualitative categories===		===Assignment of qualitative categories===

Kirilas at 09:52, 25 May 2016

2016-05-25T09:52:54Z

← Older revision		Revision as of 09:52, 25 May 2016
Line 13:		Line 13:
	* Performs a similarity search and displays top 5 most similar structures from the training set of the model along with their names, experimental results, and literature references.		* Performs a similarity search and displays top 5 most similar structures from the training set of the model along with their names, experimental results, and literature references.
	* Training of the model using ‘in-house’ data generated by ‘in-house’ screening protocol.		* Training of the model using ‘in-house’ data generated by ‘in-house’ screening protocol.

			<span style="color:red; font-weight: bold;">IMPORTANT NOTE:</span>

			If you installed ACD/Percepta v. 2016 as an in-place upgrade over the previous version, hERG Inhibition module will continue using the same set of Self-training libraries that was configured in your previous installation. To take advantage of the new, significantly extended built-in library that comes with the current version of the software, you will need to click "Configure" and manually select the folowing entry: ''hERG-I (Ki less than 10 uM) v. 1.3 (read-only)''.

			In case of a clean installation, this library is selected automatically, and no further action is required.
			<br />

	== Interface ==		== Interface ==

Kirilas: /* Features */

2016-05-25T08:21:55Z

Features

← Older revision		Revision as of 08:21, 25 May 2016
Line 8:		Line 8:

	===Features===		===Features===
	* The binomial QSAR model utilizes a training set of more than 600 compounds with experimental results mainly collected from the published hERG inhibition studies by patch-clamp method.
	* Predicts the probability for a compound to inhibit hERG channel at clinically relevant concentrations (K<sub>i</sub> < 10 μM).		* Predicts the probability for a compound to inhibit hERG channel at clinically relevant concentrations (K<sub>i</sub> < 10 μM).
	* Calculates Reliability Index (RI values) of predictions that ~~indicate~~ whether tested compounds belong to Applicability Domain of predictive model.		* The predictive model is based on a data set of more than 6500 compounds with experimental results collected from published hERG inhibition studies utilizing either patch-clamp or competitive binding methods.
			* Calculates Reliability Index (RI values) of predictions that indicates whether tested compounds belong to Applicability Domain of predictive model.
	* Performs a similarity search and displays top 5 most similar structures from the training set of the model along with their names, experimental results, and literature references.		* Performs a similarity search and displays top 5 most similar structures from the training set of the model along with their names, experimental results, and literature references.
	* Training of the model using ‘in-house’ data generated by ‘in-house’ screening protocol.		* Training of the model using ‘in-house’ data generated by ‘in-house’ screening protocol.
	~~<br />~~

	== Interface ==		== Interface ==

Kirilas: /* Model features & prediction accuracy */

2013-02-14T11:00:55Z

Model features & prediction accuracy

← Older revision		Revision as of 11:00, 14 February 2013
Line 51:		Line 51:

	===Model features & prediction accuracy===		===Model features & prediction accuracy===
	The model was ~~developed with Algorithm Builder~~ using ~~a novel methodology consisting of two parts:~~		The predictive model of hERG inhibition was derived using GALAS (Global, Adjusted Locally According to Similarity) modeling methodology (please refer to [http://www.ncbi.nlm.nih.gov/pubmed/20373217] for more details).
	* Global ~~baseline statistical model employing binomial PLS with multiple bootstrapping using a predefined set of fragmental descriptors.~~
	* Local correction to ~~baseline prediction based on analysis of experimental data for similar compounds.~~
	~~The underlying~~ methodology ~~enables obtaining an intrinsic evaluation of prediction confidence by the means of Reliability Index~~ (~~RI) values calculated for each prediction. RI ranging from 0~~ to ~~1 serves as an indication whether a submitted compound falls within the Model Applicability Domain~~. ~~Two criteria influence the calculation od Reliability Index of a prediction:~~
	* Similarity of the analyzed molecule to compounds in the Self-training Library (prediction is unreliable if no similar compounds have been found in the Library).
	* Consistency of experimental data for similar compounds (discrepant data for similar molecules, i.e. ~~alternating hERG blockers and hERG non-blockers lead to lower RI values~~).

	The ~~presented~~ method also ~~froms~~ the basis of model Trainability. 'Trainable model' methodology addresses the issue of the chemical space of ‘in-house’ libraries being considerably wider than that of publicly available data which results in limited applicability of most third-party QSARs for analysis of ‘in-house’ data. The ‘Training engine‘ makes appropriate corrections for systematic deviations produced by the baseline QSAR model based on analysis of similar compounds from the experimental data library. Expansion of this Self-training Library with user-defined experimental data for new compounds leads to instant improvement of prediction accuracy for the respective compound classes. Moreover, addition of 'in-house' data allows adapting the existing model to the particular experimental protocol used in your company and avoiding potential issues related to discrepancies between different experimental methods used for determination of drug interactions with hERG (see [[hERG_Inhibition#Model_Trainability_Demonstration\|Model Trainability Demonstration]]) section.		Each GALAS model consists of two parts:
			* Global baseline statistical model employing binomial PLS with multiple bootstrapping using a predefined set of fragmental descriptors, that reflects general trends in mutagenicity.
			* Similarity-based routine that performs local correction of baseline predictions taking into account the differences between baseline and experimental values for the most similar training set compounds.
			<br>
			GALAS methodology also provides the basis for estimating reliability of predictions by the means of calculated Reliability Index (''RI'') value ranging from 0 to 1 that takes into account the following two criteria:
			* Similarity of tested compound to the training set molecules (prediction is unreliable if no similar compounds have been found).
			* Consistence of experimental values and baseline model prediction for the most similar similar compounds from the training set (discrepant data for similar molecules, i.e. alternating hERG blockers and hERG non-blockers lead to lower ''RI'' values).

			The used method also provides the basis of model Trainability. 'Trainable model' methodology addresses the issue of the chemical space of ‘in-house’ libraries being considerably wider than that of publicly available data which results in limited applicability of most third-party QSARs for analysis of ‘in-house’ data. The ‘Training engine‘ makes appropriate corrections for systematic deviations produced by the baseline QSAR model based on analysis of similar compounds from the experimental data library. Expansion of this Self-training Library with user-defined experimental data for new compounds leads to instant improvement of prediction accuracy for the respective compound classes. Moreover, addition of 'in-house' data allows adapting the existing model to the particular experimental protocol used in your company and avoiding potential issues related to discrepancies between different experimental methods used for determination of drug interactions with hERG (see [[hERG_Inhibition#Model_Trainability_Demonstration\|Model Trainability Demonstration]]) section.

	The accuracy of predictions for compounds within model Applicability Domain (indicated by Reliability Index values) is comparable to screening results. Predictions that are not reliable, may be instantly improved by addition of experimental data for a few similar compounds to the model Self-training Library.		The accuracy of predictions for compounds within model Applicability Domain (indicated by Reliability Index values) is comparable to screening results. Predictions that are not reliable, may be instantly improved by addition of experimental data for a few similar compounds to the model Self-training Library.

	The table below shows performance of the model on the internal validation set consisting of 151 molecules. Predictions for 103 compounds (68.2% of the validation set) within Model Applicability Domain (indicated by Reliability Index (RI) value > 0.3) are highly accurate:		The table below shows performance of the model on the internal validation set consisting of 151 molecules. Predictions for 103 compounds (68.2% of the validation set) within Model Applicability Domain (indicated by Reliability Index (''RI'') value > 0.3) are highly accurate:

	{\| class="wikitable"		{\| class="wikitable"