When data sharing gets close to 100%: what human paleogenetics can teach the open science movement

Abstract

This study analyzes data sharing regarding mitochondrial, Y chromosomal and autosomal polymorphisms in a total of 162 papers on ancient human DNA published between 1988 and 2013. The estimated sharing rate was not far from totality (97.6% ± 2.1%) and substantially higher than observed in other fields of genetic research (evolutionary, medical and forensic genetics). Both a questionnaire-based survey and the examination of Journals' editorial policies suggest that this high sharing rate cannot be simply explained by the need to comply with stakeholders requests. Most data were made available through body text, but the use of primary databases increased in coincidence with the introduction of complete mitochondrial and next-generation sequencing methods. Our study highlights three important aspects. First, our results imply that researchers' awareness of the importance of openness and transparency for scientific progress may complement stakeholders' policies in achieving very high sharing rates. Second, widespread data sharing does not necessarily coincide with a prevalent use of practices which maximize data findability, accessibility, useability and preservation. A detailed look at the different ways in which data are released can be very useful to detect failures to adopt the best sharing modalities and understand how to correct them. Third and finally, the case of human paleogenetics tells us that a widespread awareness of the importance of Open Science may be important to build reliable scientific practices even in the presence of complex experimental challenges.

Fig 1. Cumulative distributions of papers on ancient human DNA from 1988 to 2013 according to the genetic system investigated.

Fig 2. Results of the questionnaire-based survey.

Rates of responses to the question “Focusing on your overall publication experience, what is the contribution of the following factors to your choice of sharing ancient human DNA data?”. The absolute values are given in parentheses. See Materials and Methods for complete statements.

Fig 3. Sharing rates in papers concerning mitochondrial and Y chromosomal polymorphisms in humans.

Vertical bars indicate 95% confidence intervals. The total number of scrutinized datasets for each field of research is reported in parentheses. All papers were indexed in Medline from 1/1/2008 to 31/12/2011.

Fig 4. Results of the questionnaire-based survey.

Rates of responses to the question “What is the contribution of the following factors to the higher rate of data sharing in DNA studies of ancient compared to extant humans?”. The absolute values are given in parentheses. See Materials and Methods for complete statements.

Fig 5. Frequencies of sharing modalities in the four genetic research fields analyzed.

Rates of usage of different sharing modalities based on the inspection of papers indexed in Medline from 01/01/2008 to 31/12/2011. The total number of scrutinized datasets for each field of research is reported in parentheses. It should be noted that the modality “will provide on request” was observed only by Milia et al. [32].