Aquatic Toxicity LC50 - Revision history

Kirilas: Fixed broken links

2022-10-20T08:08:18Z

Fixed broken links

← Older revision		Revision as of 08:08, 20 October 2022
Line 98:		Line 98:
	\|}		\|}

	For more information regarding the modeling principles and validation results please refer to [http://~~www~~.acdlabs.com/~~download~~/~~publ/2011/acss2011_qsar~~.pdf].		For more information regarding the modeling principles and validation results please refer to [http://perceptahelp.acdlabs.com/docs/Aquatic_Tox.pdf].

Kirilas at 09:57, 15 June 2017

2017-06-15T09:57:41Z

← Older revision		Revision as of 09:57, 15 June 2017
Line 30:		Line 30:


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

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

Kirilas: /* Model features & prediction accuracy */

2016-05-25T08:06:31Z

Model features & prediction accuracy

← Older revision		Revision as of 08:06, 25 May 2016
Line 98:		Line 98:
	\|}		\|}

	For more information regarding the modeling principles and validation results ~~plase~~ refer to [http://www.acdlabs.com/download/publ/2011/acss2011_qsar.pdf].		For more information regarding the modeling principles and validation results please refer to [http://www.acdlabs.com/download/publ/2011/acss2011_qsar.pdf].

Kirilas at 08:05, 25 May 2016

2016-05-25T08:05:22Z

← Older revision		Revision as of 08:05, 25 May 2016
Line 2:		Line 2:
	<br />		<br />

	Aquatic toxicity module provides the researcher with an accurate and reliable predictive tool that may serve as a valuable first estimate of fish and ~~daphnid~~ toxicity of new chemical entities that is required under REACH. It may therefore be used as an initial screen that could compete and become at least a partial replacement of time and resource consuming experimental determination in animals.<br />		Aquatic toxicity module provides the researcher with an accurate and reliable predictive tool that may serve as a valuable first estimate of fish, daphnid, and protozoan toxicity of new chemical entities that is required under REACH. It may therefore be used as an initial screen that could compete and become at least a partial replacement of time and resource consuming experimental determination in animals.<br />


	===Features===		===Features===
	* ~~A standard measure of aquatic~~ toxicity ~~is the concentration of the compound in water that is lethal~~ to ~~50% of exposed organisms (LC50).~~		* Calculates short-term toxicity to three species that are typically used in aquatic toxicity assays. The predicted parameters include lethal concentration (LC50, mg/L) to fathead minnows (''Pimephales promelas'') and water fleas (''Daphnia magna''), as well as inhibitory growth concentration (IGC50, mg/L) to ciliate protozoa (''Tetrahymena pyriformis'').
	* Provides the predictive models of LC50 (mg/L) for two species that are typically used in aquatic toxicity assays~~: Fathead minnow~~ (Pimephales promelas) and ~~Water flea~~ (Daphnia magna).		* The calculated toxicity values are supported by reliability indices (RI) that provide a quantitative evaluation of prediction confidence. High RI shows that the calculated value is likely to be accurate, while low RI indicates that no similar compounds with consistent data are present in the training set.
	* The calculated ~~LC50~~ values are supported by reliability indices (RI) that provide ~~an estimate of the prediction accuracy.~~		* Displays 5 most similar compounds from the training set for each considered aquatic species along with experimental toxicity values and similarities to the currently analyzed compound.
	* RI values represent a quantitative evaluation of prediction confidence. High RI shows that the calculated value is likely to be accurate, while low RI indicates that no similar compounds with consistent data are present in the training set.		* The training sets used to build the models contain experimental data on aquatic toxicity to fathead minnows for about 900 compounds, to water fleas for about 600 compounds, and to ciliate protozoa for about 1100 compounds.<br />
	* The training sets used to build the models contain experimental data on aquatic toxicity for about 900 compounds ~~in case of fathead minnows~~ and about ~~600~~ compounds ~~in case of water fleas~~.<br />
	<br />		<br />

Line 19:		Line 18:
	<br />		<br />

	# Calculations are presented in the form of a table. Each row contains dedicated "Configure" and "Train" buttons to select the training library for the particular species and to add new data to that library. Predictions are made for the ~~two~~ aquatic species most frequently used for testing - ~~Fathead~~ minnows (''Pimephales promelas'') ~~and Water~~ fleas (''Daphnia magna'')		# Calculations are presented in the form of a table. Each row contains dedicated "Configure" and "Train" buttons to select the training library for the particular species and to add new data to that library. Predictions are made for the three aquatic species most frequently used for testing - fathead minnows (''Pimephales promelas''), water fleas (''Daphnia magna''), and ciliate protozoa (''Tetrahymena pyriformis'').
	# The predicted value is LC50 of the analyzed compound ~~for~~ a given organism, expressed in mg/L.		# The predicted value is either LC50, or IGC50 of the analyzed compound to a given organism, expressed in mg/L.
	# Predictions are supported by Reliability Index values ranging from 0 to 1 that serve as an intrinsic evaluation of prediction confidence:		# Predictions are supported by Reliability Index values ranging from 0 to 1 that serve as an intrinsic evaluation of prediction confidence:
	#* RI < 0.3 – Not Reliable,		#* RI < 0.3 – Not Reliable,
Line 26:		Line 25:
	#* 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
	# Up to five most similar compounds from the training set with names, CAS numbers and experimental LC50 values.		# Up to five most similar compounds from the training set with names, CAS numbers and experimental LC50/IGC50 values.
	# Click the tab to browse the similar structures for different species.		# Click the tab to browse the similar structures for different species.
	<br />		<br />
Line 40:		Line 39:
	<br />		<br />

	~~Standard~~ measure of aquatic toxicity is the concentration of the compound in water that is lethal to 50% of exposed organisms (LC50). To obtain a linear relationship with structural properties these data were converted to logarithmic form (pLC50) for modeling, but the final prediction result is returned as an original LC50 value in mg/L.		The standard measure of aquatic toxicity is the concentration of the compound in water that is lethal to 50% of exposed organisms (LC50). In the case of protozoa, the endpoint is not lethality, but inhibition of cell growth by 50% (IGC50). To obtain a linear relationship with structural properties these data were converted to logarithmic form (pLC50/pIGC50) for modeling, but the final prediction result is returned as an original LC50/IGC50 value in mg/L.


	===Experimental data===		===Experimental data===
	Experimental data ~~that was~~ used for the development of predictive models ~~was~~ collected from EPA reference databases, as well as original publications.		Experimental data used for the development of predictive models were collected from EPA reference databases, as well as original publications.
	After thorough verification of the obtained values the final data sets contained about 900 compounds with quantitative LC50 values characterizing acute toxicity to fishes (''Pimephales promelas''), ~~and about~~ 600 compounds - to water fleas (''Daphnia magna'').		After thorough verification of the obtained values the final data sets contained about 900 compounds with quantitative LC50 values characterizing acute toxicity to fishes (''Pimephales promelas''), 600 compounds - to water fleas (''Daphnia magna''), and 1100 compounds - to ciliate protozoa (''Tetrahymena pyriformis'').


	===Model features & prediction accuracy===		===Model features & prediction accuracy===

	The predictive models ~~of LC50 for all considered species~~ were 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).		The predictive models were 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).

	Each GALAS model consists of two parts:		Each GALAS model consists of two parts:

Kirilas: /* Model features & prediction accuracy */

2013-02-14T09:53:58Z

Model features & prediction accuracy

← Older revision		Revision as of 09:53, 14 February 2013
Line 64:		Line 64:
	The resulting models are highly accurate: LC50 values for aquatic species are predicted with RMSE of 0.5-0.6 log units when only predictions of moderate and high reliability (''RI'' >= 0.5) are considered .''RI'' values in the high and moderate ranges are commonly obtained for 30-60% of the validation sets. Validation results also show that the accuracy of predictions is proportional to the Reliability Index, as shown in the table below for LC50 to fishes (''P. promelas''):		The resulting models are highly accurate: LC50 values for aquatic species are predicted with RMSE of 0.5-0.6 log units when only predictions of moderate and high reliability (''RI'' >= 0.5) are considered .''RI'' values in the high and moderate ranges are commonly obtained for 30-60% of the validation sets. Validation results also show that the accuracy of predictions is proportional to the Reliability Index, as shown in the table below for LC50 to fishes (''P. promelas''):

	{\| cellpadding="2" cellspacing="0" style="border-top:~~2px~~; border-bottom:~~2px~~"		{\| cellpadding="2" cellspacing="0" style="border-top:1px solid black; border-bottom:1px solid black"
	~~\|+ <b>Table 1.</b> ACD/LogS<sub>0</sub> model performance statistics for various fractions of the internal validation set.~~
	\|-		\|-
	! style="border-bottom:1px; background:#EAEAEA" width="150" \| Subset		! style="border-bottom:1px; background:#EAEAEA" width="150" \| Subset
Line 72:		Line 71:
	! style="border-bottom:1px; background:#EAEAEA" width="100" \| <i>RMSE</i>		! style="border-bottom:1px; background:#EAEAEA" width="100" \| <i>RMSE</i>
	\|-		\|-
	\| align="center~~" height="60~~" \| ''RI'' > 0.3		\| align="center" \| ''RI'' > 0.3

	\| align="center" \|		\| align="center" \|
	{\| cellpadding="0" cellspacing="0" style="width:80%; height:40px"		{\| cellpadding="2" cellspacing="0" style="width:80%; height:40px"
	\| style="~~color:white;~~ background:#B9CDE5" align="right" width="85.7%" \| '''85.7%''' \|\| style="background:#EDF2F9" width="14.3%" \|		\| style="background:#B9CDE5" align="right" width="85.7%" \| '''85.7%''' \|\| style="background:#EDF2F9" width="14.3%" \|
	\|}		\|}

	\| align="center" \| 0.656 \|\| align="center" \| 0.797		\| align="center" \| 0.656 \|\| align="center" \| 0.797
	\|-		\|-
	\| align="center~~" height="60~~" \| ''RI'' > 0.5		\| align="center" \| ''RI'' > 0.5

	\| align="center" \|		\| align="center" \|
	{\| cellpadding="0" cellspacing="0" style="width:80%; height:40px"		{\| cellpadding="2" cellspacing="0" style="width:80%; height:40px"
	\| style="~~color:white;~~ background:#B9CDE5" align="right" width="59.4%" \| '''59.4%''' \|\| style="background:#EDF2F9" width="41.6%" \|		\| style="background:#B9CDE5" align="right" width="59.4%" \| '''59.4%''' \|\| style="background:#EDF2F9" width="41.6%" \|
	\|}		\|}

	\| align="center" \| 0.795 \|\| align="center" \| 0.501		\| align="center" \| 0.795 \|\| align="center" \| 0.501
	\|-		\|-
	\| align="center~~" height="60~~" \| ''RI'' > 0.7		\| align="center" \| ''RI'' > 0.7

	\| align="center" \|		\| align="center" \|
	{\| cellpadding="0" cellspacing="0" style="width:80%; height:40px"		{\| cellpadding="2" cellspacing="0" style="width:80%; height:40px"
	\| style="~~color:white;~~ background:#B9CDE5" align="right" width="28.0%" \| '''28.0%''' \|\| style="background:#EDF2F9" width="72.0%" \|		\| style="background:#B9CDE5" align="right" width="28.0%" \| '''28.0%''' \|\| style="background:#EDF2F9" width="72.0%" \|
	\|}		\|}

Kirilas: /* Model features & prediction accuracy */

2013-02-14T09:48:44Z

Model features & prediction accuracy

← Older revision		Revision as of 09:48, 14 February 2013
Line 49:		Line 49:

	===Model features & prediction accuracy===		===Model features & prediction accuracy===
	The models for calculating LC50 of chemicals for aquatic organisms were developed according to the same methodology as Trainable models. As a result, these models share the same advantage - possibility to obtain an intrinsic evaluation of prediction confidence by the means of Reliability Index (RI) values supporting each prediction. RI ranges from 0 to 1 and serves as an indication whether a submitted compound falls within the Model Applicability Domain:
	* RI < 0.3 – Not Reliable - compound lies outside of the Model Applicability Domain
	* RI between 0.3 and 0.5 – Borderline Reliability
	* RI between 0.5 and 0.75 – Moderate Reliability
	* RI >= 0.75 – High Reliability

	The resulting models are highly accurate: LC50 values for aquatic species are predicted with RMSE 0.5-0.6 log units when only predictions of moderate and high reliability (RI >= 0.5) are considered (RI values in the high and moderate ranges are ~~provided~~ for 30-60% of the validation sets).		The predictive models of LC50 for all considered species were 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).

			Each GALAS model consists of two parts:
			* Global (baseline) statistical model that reflects general trends in the variation of the property of interest.
			* Similarity-based routine that performs local correction of baseline predictions taking into account the differences between baseline and experimental LC50 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 that takes into account:
			* Similarity of tested compound to the training set molecules.
			* Consistence of experimental LC50 values and baseline model prediction for the most similar similar compounds from the training set.

			Reliability Index ranges from 0 to 1 (0 corresponds to a completely unreliable, and 1 - a highly reliable prediction) and serves as an indication whether a submitted compound falls within the Model Applicability Domain. Compounds obtaining predictions ''RI'' < 0.3 are considered outside of the Applicability Domain of the model.
			<br><br>
			The resulting models are highly accurate: LC50 values for aquatic species are predicted with RMSE of 0.5-0.6 log units when only predictions of moderate and high reliability (''RI'' >= 0.5) are considered .''RI'' values in the high and moderate ranges are commonly obtained for 30-60% of the validation sets. Validation results also show that the accuracy of predictions is proportional to the Reliability Index, as shown in the table below for LC50 to fishes (''P. promelas''):

			{\| cellpadding="2" cellspacing="0" style="border-top:2px; border-bottom:2px"
			\|+ <b>Table 1.</b> ACD/LogS<sub>0</sub> model performance statistics for various fractions of the internal validation set.
			\|-
			! style="border-bottom:1px; background:#EAEAEA" width="150" \| Subset
			! style="border-bottom:1px; background:#EAEAEA" width="210" \| Coverage of the entire <br> internal validation set (N=175)
			! style="border-bottom:1px; background:#EAEAEA" width="100" \| <i>R</i><sup>2</sup>
			! style="border-bottom:1px; background:#EAEAEA" width="100" \| <i>RMSE</i>
			\|-
			\| align="center" height="60" \| ''RI'' > 0.3

			\| align="center" \|
			{\| cellpadding="0" cellspacing="0" style="width:80%; height:40px"
			\| style="color:white; background:#B9CDE5" align="right" width="85.7%" \| '''85.7%''' \|\| style="background:#EDF2F9" width="14.3%" \|
			\|}

			\| align="center" \| 0.656 \|\| align="center" \| 0.797
			\|-
			\| align="center" height="60" \| ''RI'' > 0.5

			\| align="center" \|
			{\| cellpadding="0" cellspacing="0" style="width:80%; height:40px"
			\| style="color:white; background:#B9CDE5" align="right" width="59.4%" \| '''59.4%''' \|\| style="background:#EDF2F9" width="41.6%" \|
			\|}

			\| align="center" \| 0.795 \|\| align="center" \| 0.501
			\|-
			\| align="center" height="60" \| ''RI'' > 0.7

			\| align="center" \|
			{\| cellpadding="0" cellspacing="0" style="width:80%; height:40px"
			\| style="color:white; background:#B9CDE5" align="right" width="28.0%" \| '''28.0%''' \|\| style="background:#EDF2F9" width="72.0%" \|
			\|}

			\| align="center" \| 0.880 \|\| align="center" \| 0.363
			\|}

			For more information regarding the modeling principles and validation results plase refer to [http://www.acdlabs.com/download/publ/2011/acss2011_qsar.pdf].

Kristina: /* Interface */

2012-05-29T13:08:31Z

Interface

← Older revision		Revision as of 13:08, 29 May 2012
Line 18:		Line 18:
	[[Image:Aquatic_toxicity_lc50.png\|center]]		[[Image:Aquatic_toxicity_lc50.png\|center]]
	<br />		<br />


	# Calculations are presented in the form of a table. Each row contains dedicated "Configure" and "Train" buttons to select the training library for the particular species and to add new data to that library. Predictions are made for the two aquatic species most frequently used for testing - Fathead minnows (''Pimephales promelas'') and Water fleas (''Daphnia magna'')		# Calculations are presented in the form of a table. Each row contains dedicated "Configure" and "Train" buttons to select the training library for the particular species and to add new data to that library. Predictions are made for the two aquatic species most frequently used for testing - Fathead minnows (''Pimephales promelas'') and Water fleas (''Daphnia magna'')
Line 29:		Line 28:
	# Up to five most similar compounds from the training set with names, CAS numbers and experimental LC50 values.		# Up to five most similar compounds from the training set with names, CAS numbers and experimental LC50 values.
	# Click the tab to browse the similar structures for different species.		# Click the tab to browse the similar structures for different species.
			<br />


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

Kristina: /* Model features & prediction accuracy */

2012-05-29T13:07:57Z

Model features & prediction accuracy

← Older revision		Revision as of 13:07, 29 May 2012
Line 48:		Line 48:

	===Model features & prediction accuracy===		===Model features & prediction accuracy===
	The models for calculating LC50 of chemicals for aquatic organisms were developed according to the same methodology as Trainable models (e.~~g. [[hERG_Inhibition\|hERG Inhibition]], [[Ames_Genotoxicity\|Ames Genotoxicity]], etc.)~~ As a result, these models share the same advantage - possibility to obtain an intrinsic evaluation of prediction confidence by the means of Reliability Index (RI) values supporting each prediction. RI ranges from 0 to 1 and serves as an indication whether a submitted compound falls within the Model Applicability Domain:		The models for calculating LC50 of chemicals for aquatic organisms were developed according to the same methodology as Trainable models. As a result, these models share the same advantage - possibility to obtain an intrinsic evaluation of prediction confidence by the means of Reliability Index (RI) values supporting each prediction. RI ranges from 0 to 1 and serves as an indication whether a submitted compound falls within the Model Applicability Domain:
	* RI < 0.3 – Not Reliable - compound lies outside of the Model Applicability Domain		* RI < 0.3 – Not Reliable - compound lies outside of the Model Applicability Domain
	* RI between 0.3 and 0.5 – Borderline Reliability		* RI between 0.3 and 0.5 – Borderline Reliability

Kirilas: /* Interface */

2012-05-24T10:21:09Z

Interface

← Older revision		Revision as of 10:21, 24 May 2012
Line 20:		Line 20:


	# Calculations ~~performed by the predictive models~~ are presented in the form of a table. Predictions are made for the two aquatic species most frequently used for testing - Fathead minnows (''Pimephales promelas'') and Water fleas (''Daphnia magna'')		# Calculations are presented in the form of a table. Each row contains dedicated "Configure" and "Train" buttons to select the training library for the particular species and to add new data to that library. Predictions are made for the two aquatic species most frequently used for testing - Fathead minnows (''Pimephales promelas'') and Water fleas (''Daphnia magna'')
	# The predicted value is LC50 of the analyzed compound for a given organism, expressed in mg/L.		# The predicted value is LC50 of the analyzed compound for a given organism, expressed in mg/L.
	# Predictions are supported by Reliability Index values ranging from 0 to 1 that serve as an intrinsic evaluation of prediction confidence:		# Predictions are supported by Reliability Index values ranging from 0 to 1 that serve as an intrinsic evaluation of prediction confidence:
Line 28:		Line 28:
	#* RI >= 0.75 – High Reliability		#* RI >= 0.75 – High Reliability
	# Up to five most similar compounds from the training set with names, CAS numbers and experimental LC50 values.		# Up to five most similar compounds from the training set with names, CAS numbers and experimental LC50 values.
	# Click the tab to browse the similar structures for different species. ~~If no similar structures are found for a particular species, the corresponding tab is disabled. <br />~~		# Click the tab to browse the similar structures for different species.
	~~<br />~~

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

	==Technical information==		==Technical information==
	<br />		<br />

Kristina at 10:58, 15 May 2012

2012-05-15T10:58:01Z

← Older revision		Revision as of 10:58, 15 May 2012
Line 31:		Line 31:
	<br />		<br />

			<div class="mw-collapsible mw-collapsed">
	==Technical information==		==Technical information==
			<br />

			<div class="mw-collapsible-content">
	===Calculated quantitative parameters===		===Calculated quantitative parameters===
	* '''Aquatic toxicity''': standard measure of aquatic toxicity is the concentration of the compound in water that is lethal to 50% of exposed organisms (LC50). To obtain a linear relationship with structural properties these data were converted to logarithmic form (pLC50) for modeling, but the final prediction result is returned as an original LC50 value in mg/L.		<br />
	* '''Endocrine disruption''': ''In vitro'' measurement of estrogen receptor binding affinity (Log ''RBA'') estimates the relative affinity of compound to receptor compared to reference ligand estradiol: ''%RBA'' = IC50(''reference'')/IC50(''test compound'') * 100%. Here IC50 is the concentration at which the unlabeled ligand displaces half of specifically bound radiolabeled 17β-estradiol to the ER, (''reference'' estrogen in a typical experiment is the same 17β-estradiol). Experimental data were converted to binary representation with two cut-offs at Log ''RBA'' = -3, and Log ''RBA'' = 0. Predicted values are probabilities that tested compound will have Log ''RBA'' higher than the defined cut-offs. Based on the predictions compounds are classified as strong binders (Log ''RBA'' > 0), weak binders (Log ''RBA'' most probably falling in the range from -3 to 0), and non-binders (Log ''RBA'' < -3).
	* '''Irritation''': For modeling purposes experimental classification results of standard Draize test (not irritating, slightly irritating, irritating, etc.) have been transformed into a binary variable according to the following scheme: compounds producing at least moderate eye or skin irritation were considered positive and all the others - negative. The resulting probabilistic predictor estimates whether the analyzed compound is likely to act as as moderate or severe eye or skin irritant.		Standard measure of aquatic toxicity is the concentration of the compound in water that is lethal to 50% of exposed organisms (LC50). To obtain a linear relationship with structural properties these data were converted to logarithmic form (pLC50) for modeling, but the final prediction result is returned as an original LC50 value in mg/L.


	===Experimental data===		===Experimental data===
	Experimental data that was used for the development of predictive models was collected from ~~various~~ reference databases ~~(Aquatic toxicity - EPA, Endocrine disruption - FDA Endocrine Disruptors DB, Risk Assessment of Endocrine Disruptors (METI), Irritation - ECB-ESIS and RTECS)~~, as well as original publications.		Experimental data that was used for the development of predictive models was collected from EPA reference databases, as well as original publications.
	After thorough verification of the obtained values the final data sets contained:		After thorough verification of the obtained values the final data sets contained about 900 compounds with quantitative LC50 values characterizing acute toxicity to fishes (''Pimephales promelas''), and about 600 compounds - to water fleas (''Daphnia magna'').
	* About 900 compounds with quantitative LC50 values characterizing acute toxicity to fishes (''Pimephales promelas''), and about 600 compounds - to water fleas (''Daphnia magna'').
	* Nearly 1500 compounds with experimentally measured ER alpha binding affinities.
	* More than 2100 molecules in both eye and skin irritation data sets that include qualitative irritation categories determined after application of test compounds to adult albino rabbits.

	===Model features & prediction accuracy===		===Model features & prediction accuracy===
	The models for calculating LC50 of chemicals for aquatic organisms were developed according to the same methodology as Trainable ~~Tox Boxes~~ models (e.g. [[hERG_Inhibition\|hERG Inhibition]], [[Ames_Genotoxicity\|Ames Genotoxicity]], etc.) As a result, these models share the same advantage - possibility to obtain an intrinsic evaluation of prediction confidence by the means of Reliability Index (RI) values supporting each prediction. RI ranges from 0 to 1 and serves as an indication whether a submitted compound falls within the Model Applicability Domain:		The models for calculating LC50 of chemicals for aquatic organisms were developed according to the same methodology as Trainable models (e.g. [[hERG_Inhibition\|hERG Inhibition]], [[Ames_Genotoxicity\|Ames Genotoxicity]], etc.) As a result, these models share the same advantage - possibility to obtain an intrinsic evaluation of prediction confidence by the means of Reliability Index (RI) values supporting each prediction. RI ranges from 0 to 1 and serves as an indication whether a submitted compound falls within the Model Applicability Domain:
	* RI < 0.3 – Not Reliable - compound lies outside of the Model Applicability Domain		* RI < 0.3 – Not Reliable - compound lies outside of the Model Applicability Domain
	* RI between 0.3 and 0.5 – Borderline Reliability		* RI between 0.3 and 0.5 – Borderline Reliability
Line 53:		Line 54:
	* RI >= 0.75 – High Reliability		* RI >= 0.75 – High Reliability

	The predictive models of Endocrine System Disruption and Irritation potential were built using binomial PLS method in Algorithm Builder. The models incorporated essential physicochemical properties of chemicals such as ionization and molecular size as well as fragmental descriptors including predefined substructures representing structural features known to have a profound influence on the analyzed property.		The resulting models are highly accurate: LC50 values for aquatic species are predicted with RMSE 0.5-0.6 log units when only predictions of moderate and high reliability (RI >= 0.5) are considered (RI values in the high and moderate ranges are provided for 30-60% of the validation sets).

	The resulting models are highly accurate:
	* LC50 values for aquatic species are predicted with RMSE 0.5-0.6 log units when only predictions of moderate and high reliability (RI >= 0.5) are considered (RI values in the high and moderate ranges are provided for 30-60% of the validation sets).
	* Overall accuracy of ER alpha affinity predictions exceeds 85% in both training and test sets in case of general binding model (Log ''RBA'' > -3), and exceeds 90% in case of strong binding (Log ''RBA'' > 0).
	* Models for the prediction of rabbit eye and skin irritation produced overall accuracy of 78% and 73% respectively.

← Older revision		Revision as of 09:53, 14 February 2013
Line 64:		Line 64:
	The resulting models are highly accurate: LC50 values for aquatic species are predicted with RMSE of 0.5-0.6 log units when only predictions of moderate and high reliability (''RI'' >= 0.5) are considered .''RI'' values in the high and moderate ranges are commonly obtained for 30-60% of the validation sets. Validation results also show that the accuracy of predictions is proportional to the Reliability Index, as shown in the table below for LC50 to fishes (''P. promelas''):		The resulting models are highly accurate: LC50 values for aquatic species are predicted with RMSE of 0.5-0.6 log units when only predictions of moderate and high reliability (''RI'' >= 0.5) are considered .''RI'' values in the high and moderate ranges are commonly obtained for 30-60% of the validation sets. Validation results also show that the accuracy of predictions is proportional to the Reliability Index, as shown in the table below for LC50 to fishes (''P. promelas''):

	{\| cellpadding="2" cellspacing="0" style="border-top:~~2px~~; border-bottom:~~2px~~"		{\| cellpadding="2" cellspacing="0" style="border-top:1px solid black; border-bottom:1px solid black"
	~~\|+ <b>Table 1.</b> ACD/LogS<sub>0</sub> model performance statistics for various fractions of the internal validation set.~~
	\|-		\|-
	! style="border-bottom:1px; background:#EAEAEA" width="150" \| Subset		! style="border-bottom:1px; background:#EAEAEA" width="150" \| Subset
Line 72:		Line 71:
	! style="border-bottom:1px; background:#EAEAEA" width="100" \| <i>RMSE</i>		! style="border-bottom:1px; background:#EAEAEA" width="100" \| <i>RMSE</i>
	\|-		\|-
	\| align="center~~" height="60~~" \| ''RI'' > 0.3		\| align="center" \| ''RI'' > 0.3

	\| align="center" \|		\| align="center" \|
	{\| cellpadding="0" cellspacing="0" style="width:80%; height:40px"		{\| cellpadding="2" cellspacing="0" style="width:80%; height:40px"
	\| style="~~color:white;~~ background:#B9CDE5" align="right" width="85.7%" \| '''85.7%''' \|\| style="background:#EDF2F9" width="14.3%" \|		\| style="background:#B9CDE5" align="right" width="85.7%" \| '''85.7%''' \|\| style="background:#EDF2F9" width="14.3%" \|
	\|}		\|}

	\| align="center" \| 0.656 \|\| align="center" \| 0.797		\| align="center" \| 0.656 \|\| align="center" \| 0.797
	\|-		\|-
	\| align="center~~" height="60~~" \| ''RI'' > 0.5		\| align="center" \| ''RI'' > 0.5

	\| align="center" \|		\| align="center" \|
	{\| cellpadding="0" cellspacing="0" style="width:80%; height:40px"		{\| cellpadding="2" cellspacing="0" style="width:80%; height:40px"
	\| style="~~color:white;~~ background:#B9CDE5" align="right" width="59.4%" \| '''59.4%''' \|\| style="background:#EDF2F9" width="41.6%" \|		\| style="background:#B9CDE5" align="right" width="59.4%" \| '''59.4%''' \|\| style="background:#EDF2F9" width="41.6%" \|
	\|}		\|}

	\| align="center" \| 0.795 \|\| align="center" \| 0.501		\| align="center" \| 0.795 \|\| align="center" \| 0.501
	\|-		\|-
	\| align="center~~" height="60~~" \| ''RI'' > 0.7		\| align="center" \| ''RI'' > 0.7

	\| align="center" \|		\| align="center" \|
	{\| cellpadding="0" cellspacing="0" style="width:80%; height:40px"		{\| cellpadding="2" cellspacing="0" style="width:80%; height:40px"
	\| style="~~color:white;~~ background:#B9CDE5" align="right" width="28.0%" \| '''28.0%''' \|\| style="background:#EDF2F9" width="72.0%" \|		\| style="background:#B9CDE5" align="right" width="28.0%" \| '''28.0%''' \|\| style="background:#EDF2F9" width="72.0%" \|
	\|}		\|}