CNN-Peaks: ChIP-Seq peak detection pipeline using convolutional neural networks that imitate human visual inspection

Oh, Dongpin; Strattan, J. Seth; Hur, Junho K.; Bento, José; Urban, Alexander; Song, Giltae; Cherry, J. Michael

doi:10.1038/s41598-020-64655-4

Cited by 17 publications

(15 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We also tested published datasets from Oh et al, who annotated peaks and noise for H3K27ac ChIP-seq in GM12878 cells and H3K4me3 in K562 cells 26 . Performance was generally consistent with our in-house labeled data, and though MACS2 performed slightly better than LanceOtron on sensitivity, LanceOtron outperformed MACS2 on precision, selectivity, and F1 score for both the H3K27ac data (LanceOtron GM12878 H3K27ac project)( Fig.…”

Section: Resultsmentioning

confidence: 99%

“…Tools such as DeepSea 24 and Bassett 25 take genomic sequence as input and can predict regulatory genomic features with high accuracy. Proof of principle studies have also shown promise for applying these techniques to peak calling 9,26 .…”

Section: Introductionmentioning

confidence: 99%

“…Proof of principle studies have also shown promise for applying these techniques to peak calling 9,24 .…”

mentioning

confidence: 99%

See 2 more Smart Citations

LanceOtron: a deep learning peak caller for ATAC-seq, ChIP-seq, and DNase-seq

Hentges

Sergeant

Downes

et al. 2021

Preprint

View full text Add to dashboard Cite

Genomics technologies, such as ATAC-seq, ChIP-seq, and DNase-seq, have revolutionized molecular biology, generating a complete genome’s worth of signal in a single assay. Coupled with the use of genome browsers, researchers can now see and identify important DNA encoded elements as peaks in an analog signal. Despite the ease with which humans can visually identify peaks, converting these signals into meaningful genome-wide peak calls from such massive datasets requires complex analytical techniques. Current methods use statistical frameworks to identify peaks as sites of significant signal enrichment, discounting that the analog data do not follow any archetypal distribution. Recent advances in artificial intelligence have shown great promise in image recognition, on par or exceeding human ability, providing an opportunity to reimagine and improve peak calling. We present an interactive and intuitive peak calling framework, LanceOtron, built around image recognition using a wide and deep neural network. We hand-labelled 499Mb of genomic data, built 5,000 models, and tested with over 100 unique users from labs around the world. In benchmarking open chromatin, transcription factor binding, and chromatin modification datasets, LanceOtron outperforms the long-standing, gold-standard peak caller MACS2 with its increased selectivity and near perfect sensitivity. Additionally, this command-line optional approach allows researchers to easily generate optimal peak-calls using only a web interface. Together, the enhanced performance, and usability of LanceOtron will improve the reliability and reproducibility of peak calls and subsequent data analysis. This tool highlights the general utility of applying machine learning to genomic data extraction and analysis.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

LanceOtron: a deep learning peak caller for ATAC-seq, ChIP-seq, and DNase-seq

Hentges

Sergeant

Downes

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Here we compared the detection accuracy of ChIP-BIT2 to that of MACS2 (2020.4 version) [ 21 ] and CNN-Peaks [ 22 ]. MACS2 was widely used in ChIP-seq peak detection.…”

Section: Resultsmentioning

confidence: 99%

ChIP-BIT2: a software tool to detect weak binding events using a Bayesian integration approach

et al. 2021

View full text Add to dashboard Cite

Background ChIP-seq combines chromatin immunoprecipitation assays with sequencing and identifies genome-wide binding sites for DNA binding proteins. While many binding sites have strong ChIP-seq ‘peak’ observations and are well captured, there are still regions bound by proteins weakly, with a relatively low ChIP-seq signal enrichment. These weak binding sites, especially those at promoters and enhancers, are functionally important because they also regulate nearby gene expression. Yet, it remains a challenge to accurately identify weak binding sites in ChIP-seq data due to the ambiguity in differentiating these weak binding sites from the amplified background DNAs. Results ChIP-BIT2 (http://sourceforge.net/projects/chipbitc/) is a software package for ChIP-seq peak detection. ChIP-BIT2 employs a mixture model integrating protein and control ChIP-seq data and predicts strong or weak protein binding sites at promoters, enhancers, or other genomic locations. For binding sites at gene promoters, ChIP-BIT2 simultaneously predicts their target genes. ChIP-BIT2 has been validated on benchmark regions and tested using large-scale ENCODE ChIP-seq data, demonstrating its high accuracy and wide applicability. Conclusion ChIP-BIT2 is an efficient ChIP-seq peak caller. It provides a better lens to examine weak binding sites and can refine or extend the existing binding site collection, providing additional regulatory regions for decoding the mechanism of gene expression regulation.

show abstract

“…The computational results confirm the effectiveness of the indicators with 95% accuracy according to the official testing conformation report issued by G7. In addition, comparative analyses as presented in Table A1 demonstrate that our method achieves comparable performance (accuracy, precision and recall) in identifying crests and troughs of all three angular velocities than baseline models but requires quite less computation time than other baseline methods including multilayer perceptron (MLP) (Tang et al , 2015), convolutional neural network (CNN) (Oh et al , 2020), long short-term memory neural network (LSTM) (Laitala et al , 2020) and bidirectional LSTM (Bi-LSTM) (Hu et al , 2021). Actually, this advantage of computational efficiency is necessary for the implementation and adoption in real use for driving grading.…”

Section: Inertial Navigation System (Ins) Data-enabled Driving Behaviour and Road Smoothness Indicatorsmentioning

confidence: 96%

Driving performance grading and analytics: learning internal indicators and external factors from multi-source data

Zhou

Zhao

et al. 2021

IMDS

View full text Add to dashboard Cite

PurposeThe purpose of this paper is to deal with the practical challenge faced by modern logistics enterprises to accurately evaluate driving performance with high computational efficiency under the disturbance of road smoothness and to identify significantly associated performance influence factors.Design/methodology/approachThe authors cooperate with a logistics server (G7) and establish a driving grading system by constructing real-time inertial navigation data-enabled indicators for both driving behaviour (times of aggressive speed change and times of lane change) and road smoothness (average speed and average vibration times of the vehicle body).FindingsThe developed driving grading system demonstrates highly accurate evaluations in practical use. Data analytics on the constructed indicators prove the significances of both driving behaviour heterogeneity and the road smoothness effect on objective driving grading. The methodologies are validated with real-life tests on different types of vehicles, and are confirmed to be quite effective in practical tests with 95% accuracy according to prior benchmarks. Data analytics based on the grading system validate the hypotheses of the driving fatigue effect, daily traffic periods impact and transition effect. In addition, the authors empirically distinguish the impact strength of external factors (driving time, rainfall and humidity, wind speed, and air quality) on driving performance.Practical implicationsThis study has good potential for providing objective driving grading as required by the modern logistics industry to improve transparent management efficiency with real-time vehicle data.Originality/valueThis study contributes to the existing research by comprehensively measuring both road smoothness and driving performance in the driving grading system in the modern logistics industry.

show abstract

CNN-Peaks: ChIP-Seq peak detection pipeline using convolutional neural networks that imitate human visual inspection

Cited by 17 publications

References 37 publications

LanceOtron: a deep learning peak caller for ATAC-seq, ChIP-seq, and DNase-seq

LanceOtron: a deep learning peak caller for ATAC-seq, ChIP-seq, and DNase-seq

ChIP-BIT2: a software tool to detect weak binding events using a Bayesian integration approach

Driving performance grading and analytics: learning internal indicators and external factors from multi-source data

Contact Info

Product

Resources

About