Standartox: Standardizing Toxicity Data
Round 1
Reviewer 1 Report
Line 77: “Moreover, Standartox could help in reducing the millions of animals used for toxicity testing each year by facilitating access to ecotoxicity data”. There are protocols adressing this way. Example Organisation for Economic Co-operation and Development (OECD), than it shoud be mentioned.
Line 137: These initiatives partly aim for overlapping goals, 138 yet have limitations or objectives that distinguish them from Standartox:” The authors need to include the “Aggregated Computational Toxicology Resource (ACToR)” database to compare among the others databases has cited on manuscript.
The authors need to clear about Standartox database will be maintained by what institution?
The “compiled Standartox data set together with the catalog” that was exported permit the users access “and filter and aggregate” via the web application and the API, through the R package seems very interesting way.
Author Response
Response to Reviewer 1 Comments
Point 1: Line 77: “Moreover, Standartox could help in reducing the millions of animals used for toxicity testing each year by facilitating access to ecotoxicity data”. There are protocols adressing this way. Example Organisation for Economic Co-operation and Development (OECD), than it shoud be mentioned.
Response 1: We changed the sentence accordingly at line 78:
“Moreover, Standartox could help in reducing the millions of animals used for toxicity testing each year by facilitating access to ecotoxicity data, which is in favor of for example the guidelines by the Organisation for Economic Co-operation and Development (OECD) \citep{oecd_oecd_2020, hartung_chemical_2009}.”
Point 2: Line 137: These initiatives partly aim for overlapping goals, 138 yet have limitations or objectives that distinguish them from Standartox:” The authors need to include the “Aggregated Computational Toxicology Resource (ACToR)” database to compare among the others databases has cited on manuscript.
Response 2: We are thankful for pointing us to this resource. ACToR is, to our understanding a collection of various data resources and links to external databases but not a web application for filtering or aggregating ecotoxicological data per se. For this reason and because it is the underlying database for Comptox we added the following sentences to the Comptox paragraph at line 146:
“Comptox is built on the Aggregated Computational Toxicology Resource (ACToR) database, which constitutes the basis for several applications published by the EPA. It collects physicochemical and toxicological data on more than 500,000 environmental chemicals and pharmaceutical compounds from various resources and presents them in a curated list on the web \citep{judson_actor_2008, judson_aggregating_2012}. However, no filter mechanisms or aggregation methods are provided in ACToR per se.”
Point 3: The authors need to clear about Standartox database will be maintained by what institution?
Response 3: The database will be maintained by the working group Quantitative Landscape Ecology at the Institute of Environmental Sciences Landau, Germany and Andreas Scharmüller, the first author of the article.
Point 4: The “compiled Standartox data set together with the catalog” that was exported permit the users access “and filter and aggregate” via the web application and the API, through the R package seems very interesting way.
Response 4: Thank you. We are convinced that script-able and harmonized approaches to retrieve ecotoxicological data are currently lacking.
Reviewer 2 Report
The authors developed Standartox, a tool and database which can provide a single aggregated data point for a specific chemical-organism test combination. This database is based on ECOTOX database, and has addressed the limitations of current databases. The paper is well written.
There are some comments.
- From Figure 2, we can notice the various data distributions before aggregation, in which the consistence between geometric mean and the value of highest probability density is different, raising the concern about whether geometric mean is the best choice for different data distributions. I also noticed that there are still some inconsistences between Standartox and other databases in Figure 3. The authors may improve the database by indicating confidence value of the aggregated data point by considering the validity of using geometric mean for different data distributions or adding other kinds of aggregated data to show the consistence between them.
- For the processing of the database, more detail and results about quality control for transferring information from ECOTOX database to be available data for aggregation in Standartox should be added.
- In Figure 1, the superscripts of “C” and “E” are not clear due to the color.
Author Response
Response to Reviewer 2 Comments
Point 1: The manuscript describes a database which delivers data from the EPA ECOTOX database in a processed manner. Unfortunately, the link given in the manuscript led to a version which was not functioning for me. I have my personal concerns about such databases as users might be tempted to neglect the raw data and results might be used by persons who just rely on data quality standards and do not consider the original data. However, there might be persons who want to work in this way. In general the manuscript appears sound and the approach is interesting. The quality of the database will be proven in real life by the users. I therefore recommend to publish the manuscript as it is.
Response 1: Currently, we still experience errors when loading www.standartox.uni-landau.de on some web browsers on the Microsoft Windows operating systems. We work on removing those. However, we haven’t experienced or heard of any issues when using the R-package.
Regarding your concerns about the Standartox database we agree with you that Standartox is not a panacea replacing other filter and aggregation methods for ecotoxicological test data. Manually quality assessed test data definitely contribute to accurate toxicity estimations and in turn, they can also be used to improve model or averaging efforts. However we reckon that Standartox helps getting a grasp of the enormous amounts of ecotoxicological data.
Reviewer 3 Report
The manuscript describes a database which delivers data from the EPA ECOTOX database in a processed manner. Unfortunately, the link given in the manuscript led to a version which was not functioning for me. I have my personal concerns about such databases as users might be tempted to neglect the raw data and results might be used by persons who just rely on data quality standards and do not consider the original data. However, there might be persons who want to work in this way. In general the manuscript appears sound and the approach is interesting. The quality of the database will be proven in real life by the users. I therefore recommend to publish the manuscript as it is.
Author Response
Response to Reviewer 3 Comments
Point 1: From Figure 2, we can notice the various data distributions before aggregation, in which the consistence between geometric mean and the value of highest probability density is different, raising the concern about whether geometric mean is the best choice for different data distributions. I also noticed that there are still some inconsistences between Standartox and other databases in Figure 3. The authors may improve the database by indicating confidence value of the aggregated data point by considering the validity of using geometric mean for different data distributions or adding other kinds of aggregated data to show the consistence between them.
Response 1: We think that the geometric mean is the best choice for various distributions because it is, in comparison to the arithmetic mean not as strongly influenced by outliers. However, in contrast to the median the geometric mean also does not completely ignore outliers. Furthermore it is also commonly used in the derivation of Species Sensitivity Distributions (SSD). Therefore, we added the study: “Species Sensitivity Distributions for Use in Environmental Protection, Assessment, and Management of Aquatic Ecosystems for 12 386 Chemicals” describing the use of SSDs and its underlying geometric mean aggregations (Posthuma et al. 2019). The following sentence was added to Line 79:
“Posthuma et al. \citet{posthuma_species_2019} showed the usefulness of SSDs and its underlying geometric mean aggregations when assessing environmental effects of chemicals.”
Concerning the “inconsistences between Standartox and other databases”: This can be attributed to several factors, for example the missing of studies in the underlying ECOTOX database. Standartox clearly doesn’t aim for exact consistency with other databases. The comparison in Figure 3 is meant to show general coherence with other database values, not an exact prediction of each of those.
We are thankful for suggesting to add a confidence value to the Standartox output. We decided to include the geometric standard deviation in the Standartox output. We did so in Figure 2 too.
Point 2: For the processing of the database, more detail and results about quality control for transferring information from ECOTOX database to be available data for aggregation in Standartox should be added.
Response 2: We added more detail on the unit conversion checks in line 218:
“This assures that 652 of the 1237 units (95.3 \% of the data) are converted correctly. The remainder could not be converted and is removed.”
Point 3: In Figure 1, the superscripts of “C” and “E” are not clear due to the color.
Response 3: We increased the font size of the labels from 14 to 22pt and colored the text in white (on the colored background).
References
Posthuma, Leo, Jos van Gils, Michiel C. Zijp, Dik van de Meent, and Dick de Zwart. 2019. “Species Sensitivity Distributions for Use in Environmental Protection, Assessment, and Management of Aquatic Ecosystems for 12 386 Chemicals.” Environmental Toxicology and Chemistry 38 (4): 905–17. https://doi.org/10.1002/etc.4373.
Round 2
Reviewer 2 Report
The authors have addressed all my major concerns.
