Tool delves into journal experimental details

There has been plenty of excitement about publishers opening up their data to be used in new applications. The vision is that new tools will emerge that help researchers in ways that may not have been thought of by publishers and could not easily be provided by publishers themselves.

This idea seems to fit very well with the plans of Irish startup Scrazzl, which launched its first products at the end of February, having just finalised a deal to use Elsevier’s SciVerse data.

Scrazzl’s business plan is based around the information in the experimental sections of papers and has its roots in its founders’ own experiences of research. As the company’s managing director David Kavanagh, who did a PhD in cell biology, explained, ‘When we were doing our own research we realised the lack of structured data about the materials and tools used and a lack of good, qualitative information about products.’

He recognised that when an individual researcher or lab wants to purchase anything from a simple reagent to a large-scale and expensive laboratory equipment it would help to have a little insight into how it has been used in research elsewhere and what other researchers think of that product.

Scrazzl aims to address that gap by providing links between products in scholarly papers and the companies that produce them, product reviews about them and other papers reporting research using the same tools. Kavanagh likens it to the travel site TripAdvisor.

‘I was at a meeting and met a product manager at Elsevier just as they were starting to open up their APIs and we realised that the methods section of papers mentions equipment all the time,’ he explained. ‘Scientists could benefit from applications using this, but we could also make money from it. It makes sense for scientists and for the companies that supply materials and equipment and it is also scaleable and a value-add for publishers.’

The Scrazzl application pulls all the product information out of a journal paper and organises that information by company. This is supplemented with links to product descriptions and user-generated content such as product reviews. It can also link with inventory control so that a researcher can see that their lab does have a sample of, for example, a particular antibody and in which freezer it is stored.

There is also a social element to Scrazzl, making connections between people and previous uses of the product. Scientists log in to the tool and can then make connections with other users of the same products and ask questions about the products. ‘When you run into technical problems you often go back to primary literature. This makes it much easier,’ pointed out Kavanagh.

 

Scrazzl can provide detailed analytics on, for example, how products are used in different areas of the world 

The anticipated revenue from this, which is shared with the publishers that provide the data, should come from manufacturers of equipment and materials paying to have a click through from the details of their products to their websites.

‘We have spoken to a lot of companies and they are all very keen. It is very easy to measure impact and we would expect the sales conversion rates on these hits to be very high,’ said Kavanagh. ‘A very simple "buy now" button is not going to do much, but because we provide supplementary information this tools feeds into solving problems in science.’

Insight into research

This researcher-facing aspect of Scrazzl is planned for launch in the Summer. However, there is another facet to the company’s technology that is already available to laboratory suppliers.

Companies that make laboratory equipment have no easy way of knowing how that equipment is being used, without trawling manually through whatever published literature they can access. One of the Scrazzl product family launched in February meets this need by providing manufacturers with citation feeds, enabling them to keep track of and promote the ways that their products are used.

However, it can go further than that and Kavanagh is particularly excited about another aspect of the February launch – business intelligence or analytics. The matching between papers and products means that Scrazzl is able to provide companies with detailed insight into how their products are used in, for example, different geographical regions or different research areas, and how they compare with their competitors’ products.

Kavanagh anticipates that this aspect will become particularly interesting once scientists begin using the tool and Scrazzl is able to provide anonymous real-time information about which products researchers are looking at.

The Scrazzl team – including Kavanagh, CTO Paul Phillips and senior engineer Daniel Hunt  

How it works

The Scrazzl database currently contains details of 300,000 laboratory products, but the company aims to expand this. ‘There is the functionality to add any brand you want and we work with companies to get structured data and catalogues. Building this up is part of the reason for the staggered release [before the researcher-facing product launch],’ explained Kavanagh.

The process of matching these laboratory products with scientific papers is not trivial of course. No two papers are written in exactly the same way and they use very complicated language. However, Kavanagh pointed out that there are some common themes. Product names and the names of their suppliers tend to be mentioned in the same sentence. And there is a limit to the number of possible sentence structures. For example, a paper might say ‘Analysis was carried out using Instrument X from Manufacturer Y’ or ‘Manufacturer Y’s Product X was diluted to a concentration of ...’

The Scrazzl developers have created filters to identify such pairings automatically and then verification is done manually. ‘We can guarantee greater than 95 per cent accuracy on matching,’ said Kavanagh. This is helped by the level of detail that Scrazzl requests from suppliers. ‘We are very specific about how we receive company data,’ he explained. ‘Product name, company and catalogue number are not enough. We need to know everything that could be used to identify products.’

It is still early days and the company is still small – seven people at the last count and most of those have joined the company since a funding round in August 2011 – but Scrazzl has big plans.

Firstly, the company is working on building further publisher partnerships. ‘Elsevier is the first step, but ultimately we want cooperation with every publisher and are actively seeking other partnerships,’ said Kavanagh.

Then there are developments planned in what information the company can unearth from published experimental details. ‘We have been quite focused on one-to-one matching, but the next stage is relationships,’ said Kavanagh. ‘We want to unlock how best to use products, for example by providing details of the dilution used with a particular antibody in previous experiments. This information is not usually in the same sentence as the product name, but is very useful to researchers,’ he added. ‘And I’m sure that as we move forward there will be other opportunities too.’