Prediction of Bacterial sRNAs Using Sequence-Derived Features and Machine Learning

Small ribonucleic acid (sRNA) sequences are 50–500 nucleotide long, noncoding RNA (ncRNA) sequences that play an important role in regulating transcription and translation within a bacterial cell. As such, identifying sRNA sequences within an organism’s genome is essential to understand the impact of the RNA molecules on cellular processes. Recently, numerous machine learning models have been applied to predict sRNAs within bacterial genomes. In this study, we considered the sRNA prediction as an imbalanced binary classification problem to distinguish minor positive sRNAs from major negative ones within imbalanced data and then performed a comparative study with six learning algorithms and seven assessment metrics. First, we collected numerical feature groups extracted from known sRNAs previously identified in Salmonella typhimurium LT2 (SLT2) and Escherichia coli K12 (E. coli K12) genomes. Second, as a preliminary study, we characterized the sRNA-size distribution with the conformity test for Benford’s law. Third, we applied six traditional classification algorithms to sRNA features and assessed classification performance with seven metrics, varying positive-to-negative instance ratios, and utilizing stratified 10-fold cross-validation. We revisited important individual features and feature groups and found that classification with combined features perform better than with either an individual feature or a single feature group in terms of Area Under Precision-Recall curve (AUPR). We reconfirmed that AUPR properly measures classification performance on imbalanced data with varying imbalance ratios, which is consistent with previous studies on classification metrics for imbalanced data. Overall, eXtreme Gradient Boosting (XGBoost), even without exploiting optimal hyperparameter values, performed better than the other five algorithms with specific optimal parameter settings. As a future work, we plan to extend XGBoost further to a large amount of published sRNAs in bacterial genomes and compare its classification performance with recent machine learning models’ performance.

Prioritizing bona fide bacterial small RNAs with machine learning classifiers

Bacterial small (sRNAs) are involved in the control of several cellular processes. Hundreds of putative sRNAs have been identified in many bacterial species through RNA sequencing. The existence of putative sRNAs is usually validated by Northern blot analysis. However, the large amount of novel putative sRNAs reported in the literature makes it impractical to validate each of them in the wet lab. In this work, we applied five machine learning approaches to construct twenty models to discriminate bona fide sRNAs from random genomic sequences in five bacterial species. Sequences were represented using seven features including free energy of their predicted secondary structure, their distances to the closest predicted promoter site and Rho-independent terminator, and their distance to the closest open reading frames (ORFs). To automatically calculate these features, we developed an sRNA Characterization Pipeline (sRNACharP). All seven features used in the classification task contributed positively to the performance of the predictive models. The best performing model obtained a median precision of 100% at 10% recall and of 64% at 40% recall across all five bacterial species, and it outperformed previous published approaches on two benchmark datasets in terms of precision and recall. Our results indicate that even though there is limited sRNA sequence conservation across different bacterial species, there are intrinsic features in the genomic context of sRNAs that are conserved across taxa. We show that these features are utilized by machine learning approaches to learn a species-independent model to prioritize bona fide bacterial sRNAs.

Computational small RNA prediction in bacteria

Bacterial, small RNAs were once regarded as potent regulators of gene expression and are now being considered as essential for their diversified roles. Many small RNAs are now reported to have a wide array of regulatory functions, ranging from environmental sensing to pathogenesis. Traditionally, noncoding transcripts were rarely detected by means of genetic screens. However, the availability of approximately 2200 prokaryotic genome sequences in public databases facilitates the efficient computational search of those molecules, followed by experimental validation. In principle, the following four major computational methods were applied for the prediction of sRNA locations from bacterial genome sequences: (1) comparative genomics, (2) secondary structure and thermodynamic stability, (3) ‘Orphan’ transcriptional signals and (4) ab initio methods regardless of sequence or structure similarity; most of these tools were applied to locate the putative genomic sRNA locations followed by experimental validation of those transcripts. Therefore, computational screening has simplified the sRNA identification process in bacteria. In this review, a plethora of small RNA prediction methods and tools that have been reported in the past decade are discussed comprehensively and assessed based on their attributes, compatibility, and their prediction accuracy.