Precise Prediction of Promoter Strength Based on a De Novo Synthetic Promoter Library Coupled with Machine Learning

Zhao, Mei; Yuan, Zhenqi; Wu, Longtao; Zhou, Shenghu; Deng, Yu

doi:10.1021/acssynbio.1c00117

Cited by 43 publications

(63 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our experiments are mainly based on the Trc synthetic promoter library created by Zhao et al [4], which consisted of 3665 different synthetic promoters and used Log Fluorescence/OD 600 to indicate promoter strength. Zhao et al [4] experimented the results of LSTM, RF, XGBoost, GBDT and other models in the data. However, they did not set a separate test set, and they did not eliminate promoters with large differences (these promoters could not even be aligned).…”

Section: Methodsmentioning

confidence: 99%

“…Our work is mainly based on the synthetic promoter library created by Zhao et al (2021), which consisted of 3665 different synthetic promoters and used log Fluorescence/OD 600 to indicate promoter strength. After sequence alignment with P trc using MEGA (Kumar et al, 2008), the synthetic promoters with large differences were discarded, and 3567 synthetic promoters were obtained.…”

Section: Experiments Datasetmentioning

confidence: 99%

“…Finally, 22 promoters with an intensity distribution range of 196 times were selected from nearly 200 promoter mutants. Zhao et al (2021) constructed and characterized a mutant library of Trc promoters ( P trc ) using 83 rounds of mutation-construction-screening-characterization engineering cycles, and established a synthetic promoter library that consisted of 3665 different variants, displaying an intensity range of more than two orders of magnitude. Despite the availability of experimental methods, obtaining a small number of useful promoters from a random library typically necessitates time-consuming and laborious screening, and given the enormous number of all possible sequence combinations, the effective identification of these useful promoters is largely constrained by a relatively small mutation library.…”

Section: Introductionmentioning

confidence: 99%

“…Previous work has studied the effect of a large number of ML models. For example, Zhao et al [4] compared the prediction accuracy of LSTM, Random Forests, GBDT, XGBoost, etc. on the Trc synthetic promoter library, and believed that XGBoost had the best effect.…”

Section: Introductionmentioning

confidence: 99%

“…In EVMP, synthetic promoters are equivalently transformed into EVMP format data, including base promoter and k -mer mutations [16], which are input into BaseEncoder and VarEncoder respectively. In Trc synthetic promoter library constructed by Zhao et al [4], EVMP models had different extents of improvement compared with non-EVMP models, among which LSTM was improved by 63.1%, Transformer was improved by 32.2%, and the other models had effect improvement of 1.1% to 7.6%. Compared with the existing SOTA non-EVMP method, EVMP obtained better results and improved SOTA records by 18.9%.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

EVMP: Enhancing machine learning models for synthetic promoter strength prediction by Extended Vision Mutant Priority framework

Yang

Huang

2022

Preprint

View full text Add to dashboard Cite

In metabolic engineering and synthetic biology applications, promoters with appropriate strengths are critical. Promoter traits including excessive sequence length and restricted vocabulary size, are considered to impede the effect of natural language models on tasks involving genetic sequence. We propose EVMP (Extended Vision Mutant Priority framework), which enhances various machine learning models without concern of model structures. The synthetic promoter input to EVMP is split into base promoter and k-mer mutations, which are encoded by BaseEncoder and VarEncoder respectively. We used EVMP on various machine learning models for promoter strength prediction in Trc synthetic promoter library. The MAE (mean absolute error) of LSTM was reduced from 0.50 to 0.18, the MAE of the Transformer was reduced from 0.27 to 0.177, and the MAE of other models was also slightly reduced by EVMP. By virtual dataset expansion formed based on multiple different base promoters, EVMP utilizes ensemble learning and conclusively lowers MAE. Further investigation demonstrated that EVMP can improve performance of machine learning models by lowering the over-smoothing phenomenon caused by the similarity of synthetic promoters. Additionally, only 56% of the original amount of synthetic promoter dataset is needed to obtain an identical previous result when applying EVMP, which reduces the cost of creating the synthetic promoter library. Our research demonstrates that EVMP is versatile and robust for building synthetic sequence machine learning models. The source code is available at https://github.com/Tiny-Snow/EVMP.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Experiments Datasetmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

EVMP: Enhancing machine learning models for synthetic promoter strength prediction by Extended Vision Mutant Priority framework

Yang

Huang

2022

Preprint

View full text Add to dashboard Cite

show abstract

Advances in computational and experimental approaches for deciphering transcriptional regulatory networks

Moeckel,

Mouratidis,

Chantzi

et al. 2024

BioEssays

View full text Add to dashboard Cite

Understanding the influence of cis‐regulatory elements on gene regulation poses numerous challenges given complexities stemming from variations in transcription factor (TF) binding, chromatin accessibility, structural constraints, and cell‐type differences. This review discusses the role of gene regulatory networks in enhancing understanding of transcriptional regulation and covers construction methods ranging from expression‐based approaches to supervised machine learning. Additionally, key experimental methods, including MPRAs and CRISPR‐Cas9‐based screening, which have significantly contributed to understanding TF binding preferences and cis‐regulatory element functions, are explored. Lastly, the potential of machine learning and artificial intelligence to unravel cis‐regulatory logic is analyzed. These computational advances have far‐reaching implications for precision medicine, therapeutic target discovery, and the study of genetic variations in health and disease.

show abstract

Deep Learning‐Assisted Design of Novel Promoters in Escherichia coli

Wang,

Xu,

Tan

et al. 2023

Advanced Genetics

View full text Add to dashboard Cite

Deep learning (DL) approaches have the ability to accurately recognize promoter regions and predict their strength. Here, the potential for controllably designing active Escherichia coli promoter is explored by combining multiple deep learning models. First, “DRSAdesign,” which relies on a diffusion model to generate different types of novel promoters is created, followed by predicting whether they are real or fake and strength. Experimental validation showed that 45 out of 50 generated promoters are active with high diversity, but most promoters have relatively low activity. Next, “Ndesign,” which relies on generating random sequences carrying functional −35 and −10 motifs of the sigma70 promoter is introduced, and their strength is predicted using the designed DL model. The DL model is trained and validated using 200 and 50 generated promoters, and displays Pearson correlation coefficients of 0.49 and 0.43, respectively. Taking advantage of the DL models developed in this work, possible 6‐mers are predicted as key functional motifs of the sigma70 promoter, suggesting that promoter recognition and strength prediction mainly rely on the accommodation of functional motifs. This work provides DL tools to design promoters and assess their functions, paving the way for DL‐assisted metabolic engineering.

show abstract

Precise Prediction of Promoter Strength Based on a De Novo Synthetic Promoter Library Coupled with Machine Learning

Cited by 43 publications

References 55 publications

EVMP: Enhancing machine learning models for synthetic promoter strength prediction by Extended Vision Mutant Priority framework

EVMP: Enhancing machine learning models for synthetic promoter strength prediction by Extended Vision Mutant Priority framework

Advances in computational and experimental approaches for deciphering transcriptional regulatory networks

Deep Learning‐Assisted Design of Novel Promoters in Escherichia coli

Contact Info

Product

Resources

About