The perspective is entitled “Precompetitive preclinical ADME/Tox data: set it free on the web to facilitate computational model building and assist drug development“ . The title is self-explanatory in terms of what we are trying to communicate. The paper is online now and available here.

Abstract

There is an increasing availability of free and open access resources for chemists to use on the internet. Coupled with the increasing availability of Open Source software tools we are in the middle of a revolution in data availability and tools to manipulate these data. ChemSpider is a free access website for chemists built with the intention of providing a structure centric community for chemists. It was developed with the intention of aggregating and indexing available sources of chemical structures and their associated information into a single searchable repository and making it available to everybody, at no charge.

There are tens if not hundreds of chemical structure databases such as literature data, chemical vendor catalogs, molecular properties, environmental data, toxicity data, analytical data etc. and no single way to search across them.  Despite the fact that there were a large number of databases containing chemical compounds and data available online their inherent quality, accuracy and completeness was lacking in many regards. The intention with ChemSpider was to provide a platform whereby the chemistry community could contribute to cleaning up the data, improving the quality of data online and expanding the information available to include data such as reaction syntheses, analytical data, experimental properties and linking to other valuable resources. It has grown into a resource containing over 21 million unique chemical structures from over 200 data sources.

ChemSpider has enabled real time curation of the data, association of analytical data with chemical structures, real-time deposition of single or batch chemical structures (including with activity data) and transaction-based predictions of physicochemical data. The social community aspects of the system demonstrate the potential of this approach. Curation of the data continues daily and thousands of edits and depositions by members of the community have dramatically improved the quality of the data relative to other public resources for chemistry.

This presentation will provide an overview of the history of ChemSpider, the present capabilities of the platform and how it can become one of the primary foundations of the semantic web for chemistry. It will also discuss some of the present projects underway since the acquisition of ChemSpider by the Royal Society of Chemistry.

The EBI recently released the ChEMBL database. This is a much bigger resource as described at the home page for the resource here. The site states “ChEMBL is a database of ca. 500,000 bioactive compounds, their quantitative properties and bioactivities (binding constants, pharmacology and ADMET, etc). The data is abstracted and curated from the primary scientific literature and the data made available due to funding by the Wellcome Trust.” It is MUCH harder to curate larger databases and 1/2 a million records is a challenge.

I downloaded the data from the FTP site and took a browse of the data. There are definitely structures in the data file that we don’t have in ChemSpider but I found an issue with charge balance for many hundreds of records where the counterions were charged (for example, chloride or bromide) but the primary component was neutral. An example is here where the compound is named as a hydrochloride but the compound has the chloride anion. I think this likely arises from treatment with some type of standardizer so it should be a matter of changing the standardizer settings and regenerating. We deal with over 23 million compounds and have been through such issues ourselves when it comes to generation of structure images.

For an example of a rich record in ChEMBL take a look at this record showing the target, assay, activity type, value and reference all listed. ChEMBL is sure to be an invaluable reference for the Life Sciences.

A few months ago I had the opportunity to spend an hour on the phone with him after he had made such positive comments when the RSC acquired ChemSpider. We talked about Open Science, Open Source and models of business. We talked about the adventure of trying to change the world one step at a time by making our humble contributions to the world of science. By the end of our conversation I knew that when I met Warren we would be able to talk for many more hours as we shared many common views and, primarily, a want to make a difference.

Today I learned of the sad news that Warren had passed away. Despite the fact that I hadn’t yet managed to sit with Warren face to face I was immediately  saddened. My truth is that there is a specific type of shock I feel when someone younger than myself passes away. Warren and I talked about the impact of our chosen career paths on our relationships with our wives and the hours spent in front of a screen instead of spending them with those we share our lives with. We both reflected on the fact that we have given too much to the keyboard over the years driven by our need to make a difference. Warren’s hard work and superior programming skills and are paralleled by the fact that he was clearly a charitable contributor to science by giving his code away to the world and was, even based on only one phone call, a kind man.

My thoughts go out to his wife and family for his loss.

How would you represent this structure if you were to draw it as a 2D diagram?

We have been working with the WikiChem team of late to embed links from Wikipedia back to ChemSpider. I am humbled to know that our hard work to establish ChemSpider as a source of quality information has reached a level of trust such that Wikipedia now links from the ChemBoxes out to ChemSpider. The links are being updated on an on going basis at present with hundreds of new links already established and more being generated on an ongoing basis. Wikipedia User: Beetstra has written a ‘bot that is inserting ChemSpiderIDs across the database (see below) and we ARE doing rigorous checking of all of the links.This was using a file that we generated on our side showing links to Wikipedia from ChemSpider.

We will then be able to generate a list of all ChemBoxes/DrugBoxes without links from Wikipedia to ChemSpider and we will then make the links on our side, manually curating the structures, and then hand back a file to finish all linking. At this point we will have the backfile under control and we can perform ongoing updates as new compound pages are created on ChemSpider and, if we curate and find errors on Wikipedia or ChemSpider making a few manual edits is easy.

There are very dedicated teams on Wikipedia and ChemSpider carefully poring over data with their robots and eyeballs to create a linked data set of quality chemistry. It’s long, tedious AND important work. When its done we will have an expanded set of data to semantically link from RSC articles when we do markup.

There are three predicted logP values from three different algorithms (ACD/LogP, XlogP and AlogPs) as shown at the top of the figure. There is a predicted value and a database value from the EPISuite from the EPA (middle of the figure) and there is a LogP value from a publication with the link out indicated by the arrow (this datum was deposited by Egon Willighagen when he deposited the data from his publication). If you examine the list of data, both experimental and predicted, you will see a general value of  around 2.65+/- error. This should be compared with the CDK value listed in the ONS spreadsheet that gives a predicted value of 0.64. This was the primary reason that we were discussing the comparison…the values of predicted logP from CDK were different from the predicted values listed on ChemSpider for a number of examples in the spreadsheet.

Egon and I exchanged a couple of emails discussing the fact that logP predictions could be generated by a number of parties if there was a good Open Data training set available. A recent publication entitled “Calculation of Molecular Lipophilicity:State of the Art and Comparison of Log P Methods on More Than 96000 Compounds” performed a thorough analysis of different logP methods on a very large dataset. The publication is available online here. They compared “the predictive power of representative methods for one public (N = 266) and two in house datasets from Nycomed(N = 882) and Pfizer (N = 95 809). A total of 30 and 18 methods were tested for public and industrial datasets, respectively.” During the work they derived a simple equation based on the number of carbon atoms, NC, and the number of hetero atoms, NHET: log P = 1.46(±0.02) + 0.11(±0.001) NC – 0.11(±0.001) NHET. This equation was shown to outperform a large number of programs benchmarked in this study. This would certainly be easy to implement on ChemSpider and, just out of interest, applying this equation to toluene gives us a value of 2.23. Compare this with the values listed above.”

Unfortunately there doesn’t appear to be too many Open logP datasets available around for people to use as training sets. Also, with the thorough work reported in the publication above is it necessary to build yet another logP prediction algorithm? ACD/Labs have made their logP prediction software free for download (http://www.acdlabs.com/download/logp.html), the VCCLab software is available for free (http://www.vcclab.org/lab/alogps/), the EPISuite software is available for free (http://www.epa.gov/oppt/exposure/pubs/episuite.htm) and if you just want to predict a value for a compound not on ChemSpider then you can use the services here: http://www.chemspider.com/Services.aspx.

However, even though there are a lot of predictors available it still makes sense to gather data and provide it as an experimental dataset, made available as Open Data for the developers of such algorithms to ake the benefits of structural diversity and fresh data to potentially improve their models. If you have any logP data available please point me to the data to download or contact me offline to discuss. We are presently working on enhancing our data model to provide improved access to experimental data on ChemSpider as well as access to the predicted data via web services. More to follow…

“In collaboration with the German National Library of Science and Technology (TIB) Thieme is the first publisher to make primary chemistry data accessible worldwide. Analytical data, from various experiments, is the foundation of research work and scientific papers. From now on, primary data will be registered and made available online via the Thieme eJournals website (www.thieme-connect.com/ejournals) using digital object recognition in the form of Digital Object Identifiers (DOI). This will enable scientists to easily locate research articles, including accompanying data, and make enhanced use of the scientific content.”

There has been a lot of discussion over the years regarding making available “primary data”. We offered to do this on the ChemSpider Journal of Chemistry : if people wanted to submit analytical data with an article that we published then we would post them as spectra associated with the article. Unfortunately the general consensus based on a few conversations that I had is that it is a lot of work to prepare data and deposit it. This is one of the reasons that, until now, publishers have generally made the spectral data available as plots and printouts of the data. These data are generally made available as electronic supplementary data. These data ARE valuable even in that form but, and I believe that the majority of scientists would agree, they would be of more valuable if they were available in a format that would allow display in online applets, downloadable for processing and expansions etc. The RSC would certainly welcome the availability of spectral data associated with publications especially since they can now be hosted on ChemSpider.

Thieme have actually managed to pull off quite a coup and I commend them for their efforts. The first example datasets are available here. The listing includes “FIDs and associated files for the ¹H, ¹³C and DEPT NMR spectra for compounds 14, (S_S)-23, (S_S)-25, (R_S)-26, 27, (S_S)-28, (R_S,S_S)-29, 30, (R_S)-36, (S_S)-36, (S_S)-37, 38, (R_S)-39, (S_S)-39, (S_S)-44, (R_S)-46, (S_S)-46, (R_S)-48, (S_S)-48, (S_S)-49, 52, (R_S)-53, (R_S)-55, (R_S)-57, (S_S)-57, (S_S)-58, (R_S)-61, (S_S)-61, (R_S)-62, (S_S)-62, (R_S)-65 and (S_S)-65 are summarized.” That’s a lot of data.

Since these are primary data they cannot be copyrighted so I chose to download the data, take a look and insert a couple into ChemSpider as an example of what can be done with these data. The associated PDF for the data says “The files can be processed using the following programs: MestReC, Bruker’s WINNMR and XWINNMR.” The files came as binary Bruker files so needed to be reprocessed and, in order to be deposited, had to be converted to JCAMP-DX format, the format supported by the JSpecView applet used on ChemSpider to display spectra. In order to this I am fortunate to have access to ACD/NMR Processor, a product I managed for a few years while working at ACD/Labs. This product also supports the Bruker format so I imported the data, processed and exported as JCAMP and imported to ChemSpider.  For compound 14 I have attached the H1 and C13 spectra and they can be seen here. I didn’t attach the “DEPT spectrum” yet. In order for me to download the spectra, redraw the structure, process the spectra, export as JCAMP and deposit to ChemSpider took about 15 minutes. However, there are a lot of spectra and it will take me a while. There are 32 compounds, I assume 3 spectra per compound (HNMR, CNMR and DEPT) so that’s a total of 96 spectra. It’ll take me about 10-12 hours just to deposit this collection so that’s a lot of work to do in my spare time. If anyone wants to help out and can process the spectra to deposit please do!

One of the spectra are shown below using the Spectral Embed function we introduced previously:

This is a rich collection of data…it can feed the Spectral Game described in this article. I look forward to getting the data onto ChemSpider and will be following up with Thieme to see if we can work together to host the data in a more generic format for the future. It’s a shame that the data are locked into a binary file format that needs reprocessing to view and I believe display through the JSpecView applet is advantageous for all. I encourage Thieme to consider also making the structure collection available in molfile, SMILES, InChI and InChIKey format – the InChIs will make the article discoverable via internet searches and through the InChI Resolver while the download of molfiles will speed up the loading process to ChemSpider and other systems.