Systematic benchmarking of tools for CpG methylation detection from Nanopore sequencing

Yuen, Zaka Wing-Sze; Srivastava, Akanksha; Daniel, Runa; McNevin, Dennis; Jack, Cameron; Eyras, Eduardo

doi:10.1101/2020.10.14.340315

Cited by 27 publications

(53 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the training data of Simpson_Hsapiens_CG_SssI, all three models achieve performances ranked next to the best reported results of Megalodon (r=0.9860, r 2 = 0.9723, RMSE=0.0758) on the dataset (Yuen et al, 2020). BiRNN achieves the best Pearson correlation r=0.9828 and r 2 =0.9658, while refine BERT achieves minimal RMSE of 0.0732 among the evaluated three models.…”

Section: Cross-sample Evaluationmentioning

confidence: 82%

“…We compare BERT models with the state-of-the-art biRNN model, which is used as the basic network structure in DeepMOD (Liu et al, 2019) and DeepSignal (Ni et al, 2019). To compare with other non-deep-learningbased methods, we utilized the CpG benchmark pipeline (Yuen et al, 2020) as a pivot.…”

Section: Methodsmentioning

confidence: 99%

“…We then conduct the cross-sample evaluation. To compare with other nondeep-learning based methods, we utilize the benchmark pipeline (Yuen et al, 2020) as a pivot. We test models on the same benchmark dataset1, which is generated based on Simpson's E.coli dataset with different methylation levels.…”

Section: Cross-sample Evaluationmentioning

confidence: 99%

“…Each mixture contains approximately 2400 reads. More detailed information can be found in (Yuen et al, 2020).…”

Section: Cross-sample Evaluationmentioning

confidence: 99%

“…Simpson_Hsapiens_CG_SssI is sequenced by the same group on different species, while Stoiber_Ecoli_CG_SssI is sequenced by a different group on the same species. We use METEORE pipeline (Yuen et al, 2020) to generate violin plots for model predictions on each mixture. The Pearson's correlation r, coefficient of determination r 2 and root mean square error (RMSE) are used as the evaluation metrics for each model.…”

Section: Cross-sample Evaluationmentioning

confidence: 99%

See 4 more Smart Citations

On the application of BERT models for nanopore methylation detection

Zhang

Hatakeyama

Yamaguchi

et al. 2021

Preprint

View full text Add to dashboard Cite

DNA methylation is a common epigenetic modification, which is widely associated with various biological processes, such as gene expression, aging, and disease. Nanopore sequencing provides a promising methylation detection approach through monitoring abnormal signal shifts for detecting modified bases in target motif regions. Recently, model-based approaches, especially those with deep learning models, have achieved significant performance improvements on nanopore methylation detection. In this work, we explore using bidirectional encoder representations from transformers (BERT) for doing the task, which can provide non-recurrent neural structures for fast parallel computation. We find original BERT architecture does not work as well as the bidirectional recurrent neural network (biRNN) on the nanopore methylation prediction task. Through further analysis, we observe recurrent patterns of positional-signal-shift in the context window surrounding target 5-methylcytosine (5mC) and N6-methyladenine (6mA) motifs. We propose a refined BERT with relative position representation and center hidden units concatenation, which takes account of task-specific characters into modeling. We perform systematic evaluations in-sample and cross-sample. The experiment results show that the refined BERT model can achieve competitive or even better results than the state-of-the-art biRNN model, while the model inference speed is about 6x faster. Besides, on the cross-sample evaluation of datasets from the different research groups, BERT models demonstrate a good generalization performance.

show abstract

Section: Cross-sample Evaluationmentioning

confidence: 82%

Section: Methodsmentioning

confidence: 99%

Section: Cross-sample Evaluationmentioning

confidence: 99%

“…Each mixture contains approximately 2400 reads. More detailed information can be found in (Yuen et al, 2020).…”

Section: Cross-sample Evaluationmentioning

confidence: 99%

Section: Cross-sample Evaluationmentioning

confidence: 99%

See 3 more Smart Citations

On the application of BERT models for nanopore methylation detection

Zhang

Hatakeyama

Yamaguchi

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

Cell-free DNA technologies for the analysis of brain cancer

Mair

Moulière

2021

Br J Cancer

View full text Add to dashboard Cite

Survival for glioma patients has shown minimal improvement over the past 20 years. The ability to detect and monitor gliomas relies primarily upon imaging technologies that lack sensitivity and specificity, especially during the post-surgical treatment phase. Treatment-response monitoring with an effective liquid-biopsy paradigm may also provide the most facile clinical scenario for liquid-biopsy integration into brain-tumour care. Conceptually, liquid biopsy is advantageous when compared with both tissue sampling (less invasive) and imaging (more sensitive and specific), but is hampered by technical and biological problems. These problems predominantly relate to low concentrations of tumour-derived DNA in the bloodstream of glioma patients. In this review, we highlight methods by which the neuro-oncological scientific and clinical communities have attempted to circumvent this limitation. The use of novel biological, technological and computational approaches will be explored. The utility of alternate bio-fluids, tumour-guided sequencing, epigenomic and fragmentomic methods may eventually be leveraged to provide the biological and technological means to unlock a wide range of clinical applications for liquid biopsy in glioma.

show abstract

Systematic benchmarking of tools for CpG methylation detection from nanopore sequencing

et al. 2021

Self Cite

View full text Add to dashboard Cite

DNA methylation plays a fundamental role in the control of gene expression and genome integrity. Although there are multiple tools that enable its detection from Nanopore sequencing, their accuracy remains largely unknown. Here, we present a systematic benchmarking of tools for the detection of CpG methylation from Nanopore sequencing using individual reads, control mixtures of methylated and unmethylated reads, and bisulfite sequencing. We found that tools have a tradeoff between false positives and false negatives and present a high dispersion with respect to the expected methylation frequency values. We described various strategies to improve the accuracy of these tools, including a consensus approach, METEORE (https://github.com/comprna/METEORE), based on the combination of the predictions from two or more tools that shows improved accuracy over individual tools. Snakemake pipelines are also provided for reproducibility and to enable the systematic application of our analyses to other datasets.

show abstract

Systematic benchmarking of tools for CpG methylation detection from Nanopore sequencing

Cited by 27 publications

References 27 publications

On the application of BERT models for nanopore methylation detection

On the application of BERT models for nanopore methylation detection

Cell-free DNA technologies for the analysis of brain cancer

Systematic benchmarking of tools for CpG methylation detection from nanopore sequencing

Contact Info

Product

Resources

About