Designing experiments is not an easy process. I can relate, having recently been at the bench myself before joining the BenchSci team. If your lab doesn’t already have an established set of protocols and reagents, then the next best resource is the scientific literature in your field. Not only does it take hours to comb through paper after paper to find exactly what you need, but I’m sure many scientists have experienced the frustration of finding a paper that contained the exact experiment you were looking to perform, only to find that the crucial details for the reagents you need to reproduce the experiment are absent.

At BenchSci, our machine-learning algorithm can read scientific papers and make an association between the usage of commercially available antibodies and the figure in which the results are displayed in the paper. Our goal is to help scientists find the reagents they need to perform their experiments quickly and efficiently.

We feel a great sense of accomplishment when we hear feedback from scientists saying that the BenchSci platform allowed them to find papers they were never able to find themselves, or find figures containing the antibody they needed to perform the experiment that has been holding back their progress. This is the driving force for us to constantly work on improving the platform.

To train our machine, we had to identify commonalities in how scientists report their experimental procedures and results. This required manually going through the scientific literature to characterize the writing patterns of scientists. It was a gruelling process, as there is no standard for antibody citation in the literature.

Based on our analysis, only 15-20% of papers that used antibodies reported the catalog number of the antibody product. Not only is this a major issue for scientists trying to reproduce experiments, but because the product information is incomplete, it makes it difficult for us to establish use cases linking antibody products to the figures in which they were used. Our team of scientists and engineers are currently working together to develop a strategy to overcome this issue, but what can we do moving forward?

Call to Action

Reproducibility in science is a major issue and there are several organizations that strive to create a high standard in scientific literature. Some examples of these organizations are the Global Biological Standards Institute (GBSI) and SciCrunch, who promote the use of best practices and standards in biomedical research.

What can you do to help? In relation to antibody citations in the literature, the most important thing is to cite your reagents in full. This means including: antibody name, supplier and catalog number.

Citation example: Antibody X was obtained from Supplier Y, Catalog #123

An even better citation method that is gaining momentum with scientists was created by the Research Resource Initiative. This initiative was launched with the intention of creating a resource to standardize the citation of biomedical resources based on the references in which they were generated or used.

Unique identifiers were created to cite antibodies, cell lines, organisms and tools. Each unique identifier for antibody products is formatted with the prefix RRID, and a registry ID consisting of a tag referencing the source authority, like AB indicating the antibody registry, and an ID number (ex. RRID: AB_98765). This identifier is to be cited along with supplier, catalog number and lot number.

RRID Citation Example: Supplier Y Cat#123 Lot#456 RRID: AB_98765

To get more information and/or find the unique RRID identifier for your reagent check out the Resource Identification Portal on the SciCrunch website.

As scientists, we shouldn’t underestimate the importance of being thorough. Not only would it help BenchSci find the best antibody for you, but it will also help your fellow scientists in need.

To learn more about this topic, we encourage local scientists to attend our symposium later this summer in Toronto addressing proper practices for antibody validation and citation in scientific literature. Stay informed by signing up for our monthly newsletter. Sign up here.

Sign Up and Search Antibodies Now BenchSci uses machine learning to decode  published data and  present figures with actionable, experiment-specific insights. Learn More

Written By:
Casandra Mangroo