Creating huskinDB, a Database for Human Skin Permeation

Like Comment
Read more

Estimation of human skin permeation by chemicals is a vital step in the development of new cosmetic or pharmaceutical products. It is essential for occupational exposure risk assessment. Back in 2016, as a group of researchers working with data scientific approaches in chemoinformatics, we wandered about the state of research on skin permeation prediction. After a brief inspection of the models, we turned our attention to the data sets that were used to train them. The first thing we noticed was the fact that the researchers were using different measurement units, so we decided to check whether the conversion was done properly. Then we decided to check the correctness of the values by comparing them to the ones reported in the original publications. While doing so, we discovered that there is quite a bit of useful information that we can extract and later use for our prediction models. The long hard road of compiling huskinDB, human skin permeation database, has begun.

Even though our database contains information from, what might seem, rather small number of publications (94 as of November 2020), curation of the information took the most effort and time, mainly due to the heterogenicity of reported conditions under which the data points were obtained. We were surprised by the number of unknown parameters due to, for example, some researchers not reporting the donor medium type or the skin thickness used in experimental procedures. This highlights the importance of research communication in a clearly defined, exhaustive way enabling others to reuse or reproduce the data.

Recognising that our compiled data set contains the most comprehensive collection of the publicly accessible data on skin permeation, we decided that it has to be made easily accessible and searchable for other researchers. For this reason, we created a huskinDB website (huskindb.drug-design.de). On this web page, it is possible to search for data records, visualize the data, and use similarity search to find skin permeation values of similar compounds. This way, the researchers can get the first impression of the data being stored and assess whether and in which way it can be used for one’s research. Looking backward, we can say that the process behind the website development was the most exciting and rewarding part of the project (certainly, in the same way as was writing of the publication after the local website development was finished).

In contrast to the data sets and databases reported previously, huskinDB focuses solely on human skin permeation value kp with intent to facilitate the development of skin permeation models and its assessment.

Our database provides an easily accessible platform and a framework for the collection and report of human skin permeation data. We hope that the researchers will continue the exploration in this field and submit their data to our database in the structure we elaborated. This will save other researchers time and curation effort! And by saving effort, it will also save time which could be spent, for example, with our dearest ones, especially during the current tough pandemic period.

 

Read our publication here, you can also take a look at our database directly via its homepage or synapse data repository.

Dmitri Stepanov

PhD Student, Freie Universität Berlin