A sequence-based global map of regulatory activity for deciphering human genetics

Chen, Kathleen M.; Wong, Aaron K.; Troyanskaya, Olga G.; Zhou, Jian

doi:10.1038/s41588-022-01102-2

Cited by 100 publications

(134 citation statements)

References 40 publications

(67 reference statements)

Supporting

Mentioning

131

Contrasting

Order By: Relevance

“…We also observed similar levels of performance of single and multitask models across metrics (Figure 4bc, Supplementary Figure 4ab), consistent with the successful application of multitask models to the prediction of chromatin state from DNA input in bulk and single cells [13][14][15][16][17][18][19][20][21][22][23] . Using a multitask model offers much less training overhead and faster inference times than training 100s of single task models across tasks.…”

Section: In Vitro Rna Binding Prediction With Deepbindsupporting

confidence: 77%

“…This data has in turn powered machine learning methods aimed at predicting the functional readouts of these sequences such as histone marks 5 , chromatin accessibility 6 , 3D conformation 7 , and gene expression 8 . Deep learning (DL) has become especially popular in this space, and has been successfully applied to tasks such as DNA and RNA protein binding motif detection [9][10][11][12] , chromatin state prediction [13][14][15][16][17][18][19][20][21][22][23] , transcriptional activity prediction 16,[24][25][26][27] and 3D contact prediction 28,29 . Recently, complementary models have been developed to predict data from massively parallel reporter assays (MPRAs) that directly test the gene regulatory potential of candidate elements [30][31][32] .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

EUGENe: A Python toolkit for predictive analyses of regulatory sequences

Klie

Stites²,

Jores

et al. 2022

Preprint

View full text Add to dashboard Cite

Deep learning (DL) has become a popular tool to study cis-regulatory element function. Yet efforts to design software for DL analyses in genomics that are Findable, Accessible, Interoperable and Reusable (FAIR) have fallen short of fully meeting these criteria. Here we present EUGENe (Elucidating the Utility of Genomic Elements with Neural Nets), a FAIR toolkit for the analysis of labeled sets of nucleotide sequences with DL. EUGENe consists of a set of modules that empower users to execute the key functionality of a DL workflow: 1) extracting, transforming and loading sequence data from many common file formats, 2) instantiating, initializing and training diverse model architectures, and 3) evaluating and interpreting model behavior. We designed EUGENe to be simple; users can develop and deploy workflows on new or existing datasets with functions that act on two customizable Python objects, annotated sequence data (SeqData) and PyTorch models (BaseModel). The modularity and simplicity of EUGENe also make it highly extensible and we illustrate these principles through application of the toolkit to three predictive modeling tasks. First, we train and compare a set of built-in models along with a custom architecture for the accurate prediction of activities of plant promoters from STARR-seq data. Next, we apply EUGENe to an RNA binding prediction task and showcase how seminal model architectures can be retrained in EUGENe or imported from Kipoi. Finally, we train models to classify transcription factor binding by wrapping functionality from Janngu, which can efficiently extract sequences in BED file format from the human genome. We emphasize that the code used in each use case is simple, readable, and well documented (https://eugene-tools.readthedocs.io/en/latest/index.html). We believe that EUGENe represents a springboard toward a collaborative ecosystem for DL applications in genomics research. EUGENe is available for download on GitHub (https://github.com/cartercompbio/EUGENe) along with several introductory tutorials and for installation on PyPi (https://pypi.org/project/eugene-tools/).

show abstract

Section: In Vitro Rna Binding Prediction With Deepbindsupporting

confidence: 77%

Section: Introductionmentioning

confidence: 99%

EUGENe: A Python toolkit for predictive analyses of regulatory sequences

Klie

Stites²,

Jores

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Importantly, each of these tools uses a different subset of available annotation data and thus may come to a different conclusion as to which mutations should be prioritized for follow-up. Recent tools such as GWAVA [ 68 ], DeepSEA [ 69 ], and Sei [ 70 ] use machine learning classification models to prioritize non-coding mutations. For example, based on a modified random forest algorithm, GWAVA prioritized five SNPs inside the 3’UTR of the caveolin 2 gene, CAV2 [ 71 ].…”

Section: Strategies For Resolving the Gap In Functional Interpretatio...mentioning

confidence: 99%

Interpretation of the role of germline and somatic non-coding mutations in cancer: expression and chromatin conformation informed analysis

Pudjihartono

Perry

et al. 2022

Clin Epigenet

View full text Add to dashboard Cite

Background There has been extensive scrutiny of cancer driving mutations within the exome (especially amino acid altering mutations) as these are more likely to have a clear impact on protein functions, and thus on cell biology. However, this has come at the neglect of systematic identification of regulatory (non-coding) variants, which have recently been identified as putative somatic drivers and key germline risk factors for cancer development. Comprehensive understanding of non-coding mutations requires understanding their role in the disruption of regulatory elements, which then disrupt key biological functions such as gene expression. Main body We describe how advancements in sequencing technologies have led to the identification of a large number of non-coding mutations with uncharacterized biological significance. We summarize the strategies that have been developed to interpret and prioritize the biological mechanisms impacted by non-coding mutations, focusing on recent annotation of cancer non-coding variants utilizing chromatin states, eQTLs, and chromatin conformation data. Conclusion We believe that a better understanding of how to apply different regulatory data types into the study of non-coding mutations will enhance the discovery of novel mechanisms driving cancer.

show abstract

“…TF binding and chromatin accessibility DeepBind (Alipanahi et al, 2015) DeepBind (Alipanahi et al, 2015) FactorNet (Quang and Xie, 2019) BPNet (Avsec et al, 2021b) DeepSEA (Zhou and Troyanskaya, 2015) Basset (Kelley et al, 2016) ChromBPNet (Trevino et al, 2021) Basset (Kelley et al, 2016) FactorNet (Quang and Xie, 2019) DeepMEL2 (Atak et al, 2021) DeepFIGV (Hoffman et al, 2019) DeepFIGV (Hoffman et al, 2019) MPRA-DragoNN (Movva et al, 2019) DeepMEL2 (Atak et al, 2021) DeepFun (Pei et al, 2021) DNABERT (Ji et al, 2021) Gene expression Basenji (Kelley et al, 2018) Enformer (Avsec et al, 2021a) Basenji (Kelley et al, 2018) Xpresso (Agarwal and Shendure, 2020) ExPecto (Zhou et al, 2018) Xpresso (Agarwal and Shendure, 2020) Enformer (Avsec et al, 2021a) Enformer (Avsec et al, 2021a) Chromatin conformation Akita (Fudenberg et al, 2020) DeepC (Schwessinger et al, 2020) Orca (Zhou, 2022) Lan and Corces 10.3389/fnagi.2022.1027224…”

Section: Deeplift or Deepshapmentioning

confidence: 99%

Deep learning approaches for noncoding variant prioritization in neurodegenerative diseases

Lan

Corces²

2022

Front. Aging Neurosci.

View full text Add to dashboard Cite

Determining how noncoding genetic variants contribute to neurodegenerative dementias is fundamental to understanding disease pathogenesis, improving patient prognostication, and developing new clinical treatments. Next generation sequencing technologies have produced vast amounts of genomic data on cell type-specific transcription factor binding, gene expression, and three-dimensional chromatin interactions, with the promise of providing key insights into the biological mechanisms underlying disease. However, this data is highly complex, making it challenging for researchers to interpret, assimilate, and dissect. To this end, deep learning has emerged as a powerful tool for genome analysis that can capture the intricate patterns and dependencies within these large datasets. In this review, we organize and discuss the many unique model architectures, development philosophies, and interpretation methods that have emerged in the last few years with a focus on using deep learning to predict the impact of genetic variants on disease pathogenesis. We highlight both broadly-applicable genomic deep learning methods that can be fine-tuned to disease-specific contexts as well as existing neurodegenerative disease research, with an emphasis on Alzheimer’s-specific literature. We conclude with an overview of the future of the field at the intersection of neurodegeneration, genomics, and deep learning.

show abstract

A sequence-based global map of regulatory activity for deciphering human genetics

Cited by 100 publications

References 40 publications

EUGENe: A Python toolkit for predictive analyses of regulatory sequences

EUGENe: A Python toolkit for predictive analyses of regulatory sequences

Interpretation of the role of germline and somatic non-coding mutations in cancer: expression and chromatin conformation informed analysis

Deep learning approaches for noncoding variant prioritization in neurodegenerative diseases

Contact Info

Product

Resources

About