Many funding agencies now require deposition of accepted articles in a public repository such as PubMed Central.
However, a disadvantage of whole-body exposure is that possible deposition of test article on animal fur or skin may result in additional unquantified oral exposure to the animals as a result of grooming behavior.
At the article level, we perform an automatic classification of articles containing data deposition statements, in contrast with Ochsner et al. who performed a one-time manual classification in order to assess the rate of data deposition in 2008.
In some cases academic fraud has been uncovered due to publishers requiring full deposition of data related to submitted articles, which has enabled rigorous procedures to analyse the integrity [21] of that data [22, 23].
Using these approaches, assessment can be further extended to a broader array of research outputs, via services that support the deposition of outputs other than full articles, such as Dryad (for datasets), Figshare (for the results of individual experiments, figures, datasets), and Slideshare (for presentations).
While the availability of full-text is definitely a limitation, our overall study of the prevalence of data deposition statements (Section 3.3) indicates that data deposition statements are significantly more often found in the full-text of articles, compared with abstract text.
In this article, the deposition of the low melting component indium with an active substrate cooling is presented.
In total, ~6.4% of articles (N=52 932) were found to contain a data deposition statement according to our method.
This allowed us to add to the gold standard 70 articles reporting the deposition of data in repositories that are not currently covered by MEDLINE curation, such as EMBL/EBI databases.
This is evidenced by the database breakdown of articles in our test corpus according to the MEDLINE [si] field: 110 articles reported data deposition in GenBank, 50 in PDB and only 47 in GEO.
