Computational Strategies for the Identification of a Transcriptional Biomarker Panel to Sense Cellular Growth States in Bacillus subtilis

Abstract

A goal of the biotechnology industry is to be able to recognise detrimental cellular states that may lead to suboptimal or anomalous growth in a bacterial population. Our current knowledge of how different environmental treatments modulate gene regulation and bring about physiology adaptations is limited, and hence it is difficult to determine the mechanisms that lead to their effects. Patterns of gene expression, revealed using technologies such as microarrays or RNA-seq, can provide useful biomarkers of different gene regulatory states indicative of a bacterium’s physiological status. It is desirable to have only a few key genes as the biomarkers to reduce the costs of determining the transcriptional state by opening the way for methods such as quantitative RT-PCR and amplicon panels. In this paper, we used unsupervised machine learning to construct a transcriptional landscape model from condition-dependent transcriptome data, from which we have identified 10 clusters of samples with differentiated gene expression profiles and linked to different cellular growth states. Using an iterative feature elimination strategy, we identified a minimal panel of 10 biomarker genes that achieved 100% cross-validation accuracy in predicting the cluster assignment. Moreover, we designed and evaluated a variety of data processing strategies to ensure our methods were able to generate meaningful transcriptional landscape models, capturing relevant biological processes. Overall, the computational strategies introduced in this study facilitate the identification of a detailed set of relevant cellular growth states, and how to sense them using a reduced biomarker panel.

A knowledge integration strategy for the selection of a robust multi-stress biomarkers panel for Bacillus subtilis

“…To enable the discovery of relevant cellular states using unsupervised machine learning, we further processed data using the processing steps as elaborately explained in [26]. These steps included: a) filtering genes that were invariant across conditions; b) removing genes and samples related to late-stage sporulation conditions, as sporulation produces a very strong transcriptional response across a large number of genes that would mask many other cellular states we are interested in capturing; c) normalising the expression quantities by subtracting the corresponding reference conditions within each experiment; to produce a processed condition-dependent gene expression data (Figure 1a) for downstream analysis.…”

“…In our previous study [26], we proposed a Biomarker Identification Model (BIM), which was applied to the Nicolas dataset to discover biomarker panels indicative of the cellular states for B. subtilis . This BIM first discovers different cellular states introduced by a wide range of conditions using UMAP dimension reduction [32] and Leiden clustering [33] methods.…”

“…More biomarker studies in biomedical domains [22][23][24] have exploited the massive transcriptomics data by applying machine learning methods such as classification and feature selection models [25]. We recently proposed a pipeline of machine learning models to extract diverse cellular states from condition-dependent transcriptomics data and identify gene biomarkers that can classify different stress states for B. subtilis [26].…”

Disparate Effects of Two Clerodane Diterpenes of Giant Goldenrod (Solidago gigantea Ait.) on Bacillus spizizenii

“…Nicolas P. et al [65] measured the gene expression changes in B. subtilis under several conditions covering various nutrients, aerobic and anaerobic growth, the development of motility, biofilm formation, adaptation to diverse stresses, high cell density fermentation, development of competence for genetic transformation, spore formation and germination. Another paper [66] using these data transformed the high dimensional transcriptomics data into a two-dimensional map. In this way, 10 clusters were identified, representing different changes in the transcriptome.…”