The mandate of the PsychENCODE Data Analysis Core (DAC) includes the development of novel integrative methodologies to construct a coherent interpretational framework for the data emerging from the consortium. The complexity of building such a framework lies in the diversity of experimental assays and their associated confounding factors, as well as in the inherent uncertainty regarding how the various target biological components function together. As a result, any analytical and computational methods would need to capture this high dimensionality of structure in the data. While classical, parallel computation advances at an incredible pace and continues to serve the needs of the research community, our experience with the ever- increasing complexity of neuropsychiatric datasets has motivated us to also look at other promising technological avenues. Accordingly, motivated by recent developments in the field of quantum computing (QC), we herein explore the use of QC algorithms as applied to two problems of relevance to the PsychENCODE DAC: (1) the prediction of brain-specific enhancers based on variants and functional genomic assays (Aim S1; related to Aim 1 of the parent grant); and (2) the calculation of the contributions of cell types to tissue-level gene expression and to the occurrence of psychiatric disorders like schizophrenia, autism spectrum disorder and bipolar disorder (Aim S2; related to Aim 1 of the parent grant). The nascency of QC hardware technologies and the complexity of simulating quantum algorithms on classical computing resources means that our exploration will be confined to smaller, judiciously chosen datasets.Nevertheless, the work in this supplement will serve to evaluate future prospects for the use of QC algorithms and hardware in genomic analyses. We also consider two different paradigms of QC, the quantum annealer and the quantum gate model, and weigh their efficiency relative to classical computing. Finally, we will incorporate the QC and classical predictions into PsychENCODE consortium's database and online portal for visualizing the relationships between different genetic and genomic elements, and evaluate corroborating evidence for the predictions (Aim S3; related to Aim 2 of the parent grant).
The PsychENCODE consortium has conducted extensive functional genomic analyses of samples from individuals diagnosed with psychiatric disorders aim to discover the complex biological architecture that lead from genetic and epigenetic markers of disease to the observed phenotypes. To reveal this underlying structure, the consortium relies on the use of sophisticated computational methods, including machine learning techniques, implemented on cutting-edge massively parallel computing resources by the consrtium?s Data Analysis Core (DAC). However, the scale and complexity of the tasks place significant burdens on these resources, and suggest the need for exploring alternative computing hardware technologies. This supplement to the DAC parent grant evaluates the promise of the emerging field of quantum computing to speed up large-scale computations and more efficiently explore the model landscape, using a comparative analysis of classical and quantum computing algorithms applied to problems relevant to the PsychENCODE DAC: the annotation of brain-specific enhancers and the quantification of cell-type contributions to bulk tissue gene expression.
| Wang, Daifeng; Liu, Shuang; Warrell, Jonathan et al. (2018) Comprehensive functional genomic resource and integrative model for the human brain. Science 362: |
| PsychENCODE Consortium; Akbarian, Schahram; Liu, Chunyu et al. (2015) The PsychENCODE project. Nat Neurosci 18:1707-12 |
