MetaRibo-Seq measures translation in microbiomes

Fremin, Brayon J.; Sberro, Hila; Bhatt, Ami S.

doi:10.1038/s41467-020-17081-z

Cited by 35 publications

(34 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…All of these ncRNAs were found to be transcribed (reads per kilobase million [RPKM] > 10) in RNA-Seq data from Li et al ( 1 ). Of the 65 known ncRNAs, 61 (94%) produced a Ribo-Seq signal (RPKM > 10) in Ribo-Seq experiments from Li et al ( 1 ) and in Ribo-Seq of MG1655 E. coli performed in our laboratory and recently reported ( 8 ) ( Table S1 ). Widespread coding by bacterial small RNAs has been described ( 9 , 10 ).…”

Section: Observationmentioning

confidence: 85%

Structured RNA Contaminants in Bacterial Ribo-Seq

Fremin

Bhatt

2020

mSphere

Self Cite

View full text Add to dashboard Cite

Ribosome profiling (Ribo-Seq) is a powerful method to study translation in bacteria. However, Ribo-Seq signal can be observed across RNAs that one would not expect to be bound by ribosomes. For example, Escherichia coli Ribo-Seq libraries also capture reads from most noncoding RNAs (ncRNAs). While some of these ncRNAs may overlap coding regions, this alone does not explain the majority of observed signal across ncRNAs. These fragments of ncRNAs in Ribo-Seq data pass all size selection steps of the Ribo-Seq protocol and survive hours of micrococcal nuclease (MNase) treatment. In this work, we specifically focus on Ribo-Seq signal across ncRNAs and provide evidence to suggest that RNA structure, as opposed to ribosome binding, protects them from degradation and allows them to persist in the Ribo-Seq sequencing library preparation. By inspecting these “contaminant reads” in bacterial Ribo-Seq, we show that data previously disregarded in bacterial Ribo-Seq experiments may, in fact, be used to gain partial information regarding the in vivo secondary structure of ncRNAs. IMPORTANCE Structured ncRNAs are pivotal mediators of bioregulation in bacteria, and their functions are often reliant on their specific structures. Here, we first inspect Ribo-Seq reads across noncoding regions, identifying contaminant reads in these libraries. We observe that contaminant reads in bacterial Ribo-Seq experiments that are often disregarded, in fact, strongly overlap with structured regions of ncRNAs. We then perform several bioinformatic analyses to determine why these contaminant reads may persist in Ribo-Seq libraries. Finally, we highlight some structured RNA contaminants in Ribo-Seq and support the hypothesis that structures in the RNA protect them from MNase digestion. We conclude that researchers should be cautious when interpreting Ribo-Seq signal as coding without considering signal distribution. These findings also may enable us to partially resolve RNA structures, identify novel structured RNAs, and elucidate RNA structure-function relationships in bacteria at a large scale and in vivo through the reanalysis of existing Ribo-Seq data sets.

show abstract

Section: Observationmentioning

confidence: 85%

Structured RNA Contaminants in Bacterial Ribo-Seq

Fremin

Bhatt

2020

mSphere

Self Cite

View full text Add to dashboard Cite

show abstract

“…We reasoned that a more accurate set of smORF predictions would be more likely to be translated, and thus more likely to be enriched for a Ribo-Seq signal. We generated and sequenced Ribo-Seq libraries for one Bacteroides thetaiotaomicron isolate (previously published (Sberro et al, 2019)), and four metagenomic human microbiome samples using MetaRibo-Seq, a technique for metagenomic ribosome profiling (Fremin et al, 2020).…”

Section: Predicted Smorfs Are Enriched For Ribo-seq Signalmentioning

confidence: 99%

“…We repeated this analysis using published MetaRibo-Seq data generated from stool samples of four human subjects ( Fig. 2b) (Fremin et al, 2020). Given the high number of smORFs in each sample, there is increased power to detect a MetaRibo-Seq enrichment compared to the previous example when we only queried a single microbial strain (B. thetaiotaomicron).…”

Section: Predicted Smorfs Are Enriched For Ribo-seq Signalmentioning

confidence: 99%

“…Ribo-Seq datasets were used to validate that the SmORFinder annotation tool enriches for actively translated smORFs. We used previously published Ribo-seq datasets that are available through the NCBI SRA portal under the projects PRJNA540869 (Ribo-Seq of B. thetaiotaomicron isolate) and PRJNA510123 (MetaRibo-Seq of human stool samples from 4 individuals) (Fremin et al, 2020;Sberro et al, 2019). The B. thetaiotaomicron reference genome was annotated using Prodigal configured to identify smORFs.…”

Section: Validating Smorfinder With Ribo-seq Datasetsmentioning

confidence: 99%

“…We demonstrate that the deep neural networks are more sensitive and have a higher F1 score than profile HMMs when it comes to the classification of protein families that did not exist in the training set. Then we apply this annotation tool to Ribo-Seq and MetaRibo-Seq (Fremin et al, 2020) datasets, demonstrating that its predictions are enriched for actively translated genes, and that combining predictions from different models improves precision. Next, we find evidence that the deep neural networks we trained have learned to identify Shine-Dalgarno sequences, to deprioritize the wobble position in each codon, and to group codons in a way that strongly corresponds with the codon table.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Automated prediction and annotation of small proteins in microbial genomes

Bhatt

2020

Preprint

Self Cite

View full text Add to dashboard Cite

Recent work performed by Sberro et al. (2019) revealed a vast unexplored space of small proteins existing within the human microbiome. At present, these small open reading frames (smORFs) are unannotated in existing reference genomes and standard genome annotation tools are not able to accurately predict them. In this study, we introduce an annotation tool named SmORFinder that predicts small proteins based on those identified by Sberro et al. This tool combines profile Hidden Markov models (pHMMs) of each small protein family and deep learning models that may better generalize to smORF families not seen in the training set. We find that combining predictions of both pHMM and deep learning models leads to more precise smORF predictions and that these predicted smORFs are enriched for Ribo-Seq or MetaRibo-Seq translation signals. Feature importance analysis reveals that the deep learning models learned to identify Shine-Dalgarno sequences, deprioritize the wobble position in each codon, and group codons in a way that strongly corresponds to the codon synonyms found in the codon table. We perform a core genome analysis of 26 bacterial species and identify many core smORFs of unknown function. We pre-compute small protein annotations for thousands of RefSeq isolate genomes and HMP metagenomes, and we make these data available through a web portal along with other useful tools for small protein annotation and analysis. The systematic identification and annotation of those important small proteins will help researchers to expand our understanding of this exciting field of biology.

show abstract