Autism Data Science Initiative (ADSI) Publication Guidance

Navigation #contents

Dissemination of Verification Results

Timing and Publication Embargoes

Publication Permissions and Approvals

Guidance Overview #overview

The purpose of this guidance is to provide ADSI research and verification teams with a broad set of guiding principles regarding the public dissemination of verification results. These guidelines were synthesized through several Publication Working Group meetings, with input from members of the Primary Research Project (PRPs) and Verification Center (VC) teams alongside National Institutes of Health (NIH) staff. The guidelines outlined here are not NIH mandates but rather are designed to help the PRP and VC teams establish their respective Verification Protocols and publication strategies. The Verification Protocol established by each VC and PRP team will describe the dissemination strategy for each study, including anticipated PRP and VC publishing timelines, communication strategies, and shared authorship criteria. This guidance should not be used to guide general PRP-VC working team interactions, as these will be considerably more frequent with open and routine sharing of data resources, code, and relevant findings.

Back to Top

Dissemination of Verification Results #dissemination

A critical component of ADSI is that the VC findings are disseminated to the scientific community through publication in scientific journals. Potential avenues for this include research articles specific to PRP studies or broader verification package articles. In addition, editorials or commentaries accompanying publication of PRP findings may be appropriate. If a suitable publication outlet does not exist, ADSI researchers, in collaboration with NIH staff, may explore further dissemination platforms (e.g., websites, preprint servers, etc.).

PRP publications may include common language (developed by the Publication Working Group) to inform journals and the scientific community of forthcoming VC findings. The goal here is to clearly link the PRP and VC publications.
An example of Common Language Across Studies: This study was conducted as a part of the National Institutes of Health (NIH) Autism Data Science Initiative (ADSI). In addition to the methods and statistical analyses described here, research findings will undergo an independent verification process by a separate research team, which will be presented in a forthcoming publication.
If the VC team discovers a minor discrepancy (e.g., an error in code, statistical procedures, etc.), this should be communicated to the PRP team as soon as possible. If identified early enough, these discrepancies may be rectified prior to publication of PRP findings. If PRP and VC teams are unable to reach consensus on study results or findings, both groups should still publish their findings, though it is recommended that the two teams discuss potential interpretations of the differences noted prior to publication. Independent analyses that yield different results can pave the way for future avenues of research. For community-level clarity, PRP teams (either independently or in collaboration with the VC team) are encouraged to publish an editorial or commentary on any identified discrepancies that may impact the interpretation of data or conclusions drawn in PRP team publications.

Timing and Publication Embargoes #timing

Per this guidance, a PRP publication is defined as the first publicly available and thorough report of a new data finding; this may be in the form of a journal article or uploaded preprint. Conference proceedings, posters, presentations, or abstracts are not considered publication. VCs will be embargoed from publishing their work until the PRP team publishes their main study findings as outlined in the Verification Protocol or until twelve months after the funded period of performance ends, whichever comes first. If, due to the peer-review process timeline, the VC publication is slated for release before the publication of the corresponding work from the PRP team, the VC should make every effort to work with the journal to delay the publication of the VC paper until the PRP team’s paper is released. The PRP and VC teams should develop an anticipated publication timeline, which then should be presented in the Verification Protocol, with an understanding that verification work will be published on a rolling basis as PRP findings are made available. Given the scope of the ADSI studies, it is likely that several ADSI PRP teams will publish a series of manuscripts, as opposed to one ADSI PRP publication per team (e.g., data from Aim 1 may be published before data from Aims 2+). In this event, the VC team may publish any verification findings associated with data presented in published PRP manuscripts. The VC team should not publish or present* on any PRP data that have not yet been disseminated by the PRP team, even if these data are available to and processed by the VC team (e.g., if only PRP study Aim 1 has been published by the PRP team, data derived from study Aims 2+ may not be published by the VC team). The VC team may withhold publication of their validation efforts until all aims of a PRP have been completed, depending on data complexity, verification needs, and the establishment of a publishable verification unit. There are no embargoes on PRP team publishing efforts. PRP teams may publish regardless of VC team progress.

* Presentation of research activities is a routine part of science dissemination and trainee career development. Thus, VC teams may be given the opportunity to present on their ADSI-associated efforts prior to PRP team publication. In this event, the VC team should discuss any presentation of unpublished work with the PRP team prior to agreeing to present. In most cases, new data findings should not be presented by the VC team. Each PRP and VC team is encouraged to outline presentation boundaries in the development of their respective Verification Protocol.

Authorship #authorship

It is important that independence be maintained between PRP and VC teams, and that this independence be clearly communicated and maintained from project inception to publication. For most verification efforts, NIH advises that PRP and VC teams publish independently. If the PRP and VC teams are considering co-publication, they should describe this in the verification protocol so that NIH staff may review to ensure that this is the best course of action for verification dissemination.

The appropriateness of co-publication between PRP and VC teams will be assessed on a case-by-case basis. The proposed publication method will be included in the verification protocol with flexibility for changes across the verification process.
In select circumstances, findings from the VC team may be packaged within a PRP-led publication. In this event, VC team members should be included as co-authors or listed in the acknowledgements, as appropriate based on scientific contribution (see ICMJE Guidelines). When the VC is included as a co-author or in the acknowledgements, the methods and contributions of the VC team need to be clearly described in the Methods and/or Acknowledgements section. Again, the teams should clearly indicate in the publication the extent to which they conducted analyses separately and without influence from the other team. Finally, when teams are interested in co-publishing, the verification protocol should specify how the VC contributions will be acknowledged, either through authorship or acknowledgement.
If the VC believes they have identified an analytical discrepancy in reviewing the work performed by the PRP team, the VC must notify the PRP team by emailing the contact PI and Project Manager (PM). If the PRP and VC teams do not agree on the validity of the analysis, methodology, or result, co-publication may not be the best approach. Such differences can point to areas or topics that may benefit from future research. As previously noted, the PRPs and VCs are encouraged to publish an editorial or commentary on any identified differences in findings that may impact the interpretation of data or conclusions drawn in separate publications.

Publication Permissions and Approvals #approvals

The VCs need to share a draft of any manuscript derived from PRP data with the PRP team one month prior to submission to a scientific journal or preprint server via email to the contact PI and PM of the PRP. The PRPs also need to provide the VC with a copy of their draft manuscript upon submission to a peer-reviewed journal or pre-print. The PRP contact PI and PM will be responsible for sharing the VC draft with additional members of the PRP, including the Community Advisory Board, as needed and appropriate, keeping in mind the need for confidentiality of unpublished findings.

If the PRPs have comments or feedback on the manuscript prepared by the VCs, the VCs are not required to incorporate the feedback into the draft. However, the VCs should acknowledge receipt of any feedback the PRPs provide and provide a summary of how they chose to respond to the feedback.
Both the VC and PRP teams should notify the NIH when they submit a pre-print or manuscript for review so that NIH may help to promote the findings.
For manuscripts submitted to a peer-reviewed journal, both the PRPs and VC should notify the other team if there is a request for significant revisions as part of the peer-review process. This will ensure the VC can potentially amend their verification protocol as needed and the PRP team has an opportunity to consider the feedback received as part of the peer-review process.
If the reviewer of a PRP manuscript requests to see the results from the verification study, the PRP and VC teams should collaborate to make the verification data available to the reviewers as soon as feasible. The PRP is free to waive the embargo on publishing the verification results if it will aid in publishing the primary study results.
In publishing their verification findings, any data set released or withheld by the VC will be done so as governed by the Data Use Agreement (DUA) associated with the source data.

Autism Data Science Initiative (ADSI) Publication Guidance