Data Science and Sharing Team

National Institute Of Mental Health

Investigators

Linked publications, trials & patents

Abstract

For our sixth annual report, we have provided some highlights of our team's activities over the past fiscal year. Data Sharing In collaboration with Joyce Chung and the NIMH Clinical Director's Office, the DSST has recently released additional data from the Healthy Research Volunteers Protocol (NCT03304665) on OpenNeuro (https://openneuro.org/datasets/ds004215). The associated data paper has been accepted for publication. In collaboration with Audrey Thurm's group, 263 new MRI sessions from the Autism Subtypes Study are now available at the NIMH Data Archive (NDA) (https://nda.nih.gov/edit_collection.html?id=2368). DSST staff have curated and uploaded a dataset of cerebral protein synthesis (rCPS) to the OpenNeuro repository for the Section on Neuroadaptation and Protein Metabolism. The data were derived from 93 subjects in 4 separate PET studies with rCPS maps, anatomical MRI scans, and raw PET data. Data Curation DSST now provides the IRP with access to over 110,000 MRI scan sessions across 31 different datasets. We maintain a comprehensive list of these datasets on our website (http://cmn.nimh.nih.gov/dsst). Our most requested datasets are the UK Biobank, Human Connectome Project, and Adolescent Brain Cognitive Development Study (ABCD) datasets. Notably, we have now downloaded and curated the most recent Fast Track release of raw imaging data from the ABCD study, consisting of over 11,000 adolescents, some with Neuroimaging data from as many as three time points collected at two-year intervals. This dataset is currently in use by Dr. Karen Bermans NIMH group and by Dr. Dardo Tomasi from Dr. Nora Volkovs group at NIAAA, but any NIMH intramural researcher can easily access it on the NIH HPC. In collaboration with Dr. Armin Raznahan's group, we are also now curating Genotype-Tissue Expression (GTEx) datasets from the NHGRI Genomic Data Science Analysis, Visualization, and Informatics Lab-space (AnVIL). This 100 TB dataset is now accessible from the NIH HPC and available for use by the wider NIMH intramural research program. We are currently working with Dr. Stefano Marenco of the Human Brain Collection Core to secure the data-use agreements required to give them access to this unique dataset. In collaboration with Dr. Carlos Zarates group, we have curated and preprocessed two datasets of patients with Major Depressive Disorder (MPD) and healthy controls that were scanned before and after therapeutic ketamine infusion. These datasets are being analyzed by the Machine Learning Team (MLT) to build classifiers to identify MPD risk factors as well as response to therapies such as ketamine infusion. In collaboration with the Scientific and Statistical Computing Core (SSCC) and Dr. Jo Etzel at Washington University, our team curated a dataset that combines subjects from 7 different publicly available datasets for an fMRI quality control project. More information can be found on the projects OSF page (https://osf.io/qaesm). Training The DSST is continually providing ad hoc training while consulting with researchers and trainees throughout the NIH intramural program. This fiscal year, it also offered researchers five structured training opportunities, detailed below: 1. Since June 2021, the DSST has hosted a weekly series called Lunch & Learn, where a speaker presents a primer or hosts a discussion on a single topic relevant to DSST's mission. 2. In December of 2021, Dustin Moraczewski, Arshitha Basavaraj, and Jessica Dafflon served on the organizing committee for the DC chapter of the BrainHack Global yearly Hackathon. 3. In April of 2022, Eric Earl organized an event for the core Brain Imaging Data Structure (BIDS) and OpenNeuro teams. A prototype of the new BIDS validator and schema were produced. The post-sprint report is available (https://bit.ly/pdx-bids-sprint). 4. In April of 2022, Eric Earl presented on GitHub usage for good programming practices to the MEG Core (https://megcore.nih.gov/index.php?title=Club_MEG). 5. In July of 2022, Dustin Moraczewski and Arshitha Basavaraj spoke to the Noninvasive Neuromodulation Unit on best practices in open and reproducible science and data sharing. Preregistered Replication Starting in September of 2019, our team mentored a one-year post baccalaureate student, Nik Goyal, who submitted a paper to the preregistered replication track of the Royal Society of Open Science. Our replication efforts were successful and the paper was published in February of 2022 (https://doi.org/10.1098/rsos.201090). Collaborations In collaboration with Dr. Robert Innis, Gitte Knudsen, Melanie Ganz, and Cyril Pernet, our team is working to advance and encourage data-sharing in the PET community. This has entailed the creation of standard nomenclature, a new BIDS specification for formatting and sharing PET data, and a new public repository for sharing data. For more details on this project, see the NIMH annual report entitled OpenNeuroPET (https://intramural.nih.gov/search/searchview.taf?ipid=12210). In collaboration with the MLT, we aim to leverage information across multiple large datasets to improve upon individual phenotypic prediction models using multiple imaging modalities. Jessica Dafflon has extended a tool developed by Dylan Nielson (MLT) to consolidate preprocessing outputs to facilitate visualization and quality control and more easily identify outliers. This tool fills an important need in the community and has been downloaded more than 4,000 times (https://github.com/transatlantic-comppsych/fmriprep-group-report). In our collaboration with Armin Raznahan and the Section on Developmental Neurogenomics, we assisted with processing the structural brain images from almost 40,000 subjects within the UK Biobank dataset in order to examine complex relationships between genome and brain structure. In addition, we have assisted with calculating novel structural MRI metrics (e.g., local gyrification index) from the same dataset. In an external collaboration with Samuel Guay, from University of Montreal, Eric Earl has prepared a BIDS Extension Proposal for BIDS Phenotypic Data (https://osf.io/35sxv/). This work has also generated three OpenNeuro example datasets (https://bit.ly/bids-phenotype-example1, https://bit.ly/bids-phenotype-example2, and https://bit.ly/bids-phenotype-example3) and a tool for phenotypic data preparation (https://bit.ly/bids-phenotype-tool). In a collaboration with Dr. Mark Histed and the Unit on Neural Computation and Behavior, Carl Harris is assisting with a pipeline for signal extraction from two-photon datasets using the Neurodata Without Borders (NWB) data standard. Soon we will introduce an extension to the NWB API that allows for the storage of holographic photostimulation patterns. We have applied for a one-year Kavali seed grant from NWB to fund this effort. With the MLT, we are working on limb tracking and motion segmentation in video through a collaboration with Dr. Hendrikje Nienborg. We extracted animal body part positions in 3D using a pipeline built on DeepLabCut and AniPose, two open-source Python toolboxes. Using key points extracted from video, we also implemented an unsupervised learning approach that successfully differentiates between move and rest states in experimental video with high accuracy. COVID-19 Our work on COVID-19 survey data continues in collaboration with Drs. Francisco Pereira, Joyce Chung, and Melissa Brotman. Carl Harris has made substantial progress with machine learning models designed to predict longitudinal outcomes using kernel regression techniques. Data collection and analyses are ongoing.