Deep learning (DL) has shown potential to provide powerful representations of bulk RNA-seq data in cancer research. However, there is no consensus regarding the impact of design choices of DL approaches on the performance of the learned representation, including the model architecture, the training methodology and the various hyperparameters. To address this problem, we evaluate the performance of various design choices of DL representation learning methods using TCGA and DepMap pan-cancer datasets and assess their predictive power for survival and gene essentiality predictions. We demonstrate that baseline methods achieve comparable or superior performance compared to more complex models on survival predictions tasks. DL representation methods, however, are the most efficient to predict the gene essentiality of cell lines. We show that auto-encoders (AE) are consistently improved by techniques such as masking and multi-head training. Our results suggest that the impact of DL representations and of pretraining are highly task- and architecture-dependent, highlighting the need for adopting rigorous evaluation guidelines. These guidelines for robust evaluation are implemented in a pipeline made available to the research community.

Mycobacterium tuberculosis is the main causative agent of tuberculosis (TB)-an ancient yet widespread global infectious disease to which 1.6 million people lost their lives in 2021. Antimicrobial resistance (AMR) has been an ongoing crisis for decades; 4.95 million deaths were associated with antibiotic resistance in 2019. While AMR is a multi-faceted problem, drug discovery is an urgent part of the solution and is at the forefront of modern research.The landscape of drug discovery for TB has undoubtedly been transformed by the development of high-throughput gene-silencing techniques that enable interrogation of every gene in the genome, and their relative contribution to fitness, virulence, and AMR. A recent advance in this area is CRISPR interference (CRISPRi). The application of this technique to antimicrobial susceptibility testing (AST) is the subject of ongoing research in basic science.CRISPRi technology can be used in conjunction with the high-throughput SPOT-culture growth inhibition assay (HT-SPOTi) to rapidly evaluate and assess gene essentiality including non-essential, conditionally essential (by using appropriate culture conditions), and essential genes. In addition, the HT-SPOTi method can develop drug susceptibility and drug resistance profiles.This technology is further useful for drug discovery groups who have designed target-based inhibitors rationally and wish to validate the primary mechanisms of their novel compounds\' antibiotic action against the proposed target.

CHCHD4 (MIA40) is central to the functions of the mitochondrial disulfide relay system (DRS). CHCHD4 is essential and evolutionarily conserved. Previously, we have shown CHCHD4 to be a critical regulator of tumour cell growth. Here, we use integrated analysis of our genome-wide CRISPR/Cas9 and SILAC proteomic screening data to delineate mechanisms of CHCHD4 essentiality in cancer. We identify a shortlist of common essential genes/proteins regulated by CHCHD4, including subunits of complex I that are known DRS substrates, and genes/proteins involved in key metabolic pathways. Our study highlights a range of CHCHD4-regulated nuclear encoded mitochondrial genes/proteins essential for tumour cell growth.

Clostridium acetobutylicum is a solventogenic, anaerobic, gram-positive bacterium that is commonly considered the model organism for studying acetone-butanol-ethanol fermentation. The need to produce these chemicals sustainably and with a minimal impact on the environment has revived the interest in research on this bacterium. The recent development of efficient genetic tools allows to better understand the physiology of this micro-organism, aiming at improving its fermentation capacities. Knowledge about gene essentiality would guide the future genetic editing strategies and support the understanding of crucial cellular functions in this bacterium. In this work, we applied a transposon insertion site sequencing method to generate large mutant libraries containing millions of independent mutants that allowed us to identify a core group of 418 essential genes needed for in vitro development. Future research on this significant biocatalyst will be guided by the data provided in this work, which will serve as a valuable resource for the community.
OBJECTIVE: Clostridium acetobutylicum is a leading candidate to synthesize valuable compounds like three and four carbons alcohols. Its ability to convert carbohydrates into a mixture of acetone, butanol, and ethanol as well as other chemicals of interest upon genetic engineering makes it an advantageous organism for the valorization of lignocellulose-derived sugar mixtures. Since, genetic optimization depends on the fundamental insights supplied by accurate gene function assignment, gene essentiality analysis is of great interest as it can shed light on the function of many genes whose functions are still to be confirmed. The data obtained in this study will be of great value for the research community aiming to develop C. acetobutylicum as a platform organism for the production of chemicals of interest.

The identification of essential genes is a key challenge in systems and synthetic biology, particularly for engineering metabolic pathways that convert feedstocks into valuable products. Assessment of gene essentiality at a genome scale requires large and costly growth assays of knockout strains. Here we describe a strategy to predict the essentiality of metabolic genes using binary classification algorithms. The approach combines elements from genome-scale metabolic models, directed graphs, and machine learning into a predictive model that can be trained on small knockout data. We demonstrate the efficacy of this approach using the most complete metabolic model of Escherichia coli and various machine learning algorithms for binary classification.

Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/Cas9 system has enabled rapid advances in genomic engineering and transcriptional regulation. Specifically, CRISPR interference (CRISPRi) system has been used to systematically investigate the gene functions of microbial strains in a high-throughput manner. This method involves growth profiling using cells that have been transformed with the deactivated Cas9 (dCas9) and single-guide RNA (sgRNA) libraries that target individual genes. The fitness scores of each gene are calculated by measuring the abundance of individual sgRNAs during cell growth and represent gene essentiality. In this chapter, a process is described for functional genetic screening using CRISPRi at the whole-genome scale, starting from the synthesis of sgRNA libraries, construction of CRISPRi libraries, and identification of essential genes through growth profiling. The commensal bacterium Bacteroides thetaiotaomicron was used to implement the protocol. This method is expected to be applicable to a broader range of microorganisms to explore the novel phenotypic characteristics of microorganisms.

Advances in gene editing, in particular CRISPR interference (CRISPRi), have enabled depletion of essential cellular machinery to study the downstream effects on bacterial physiology. Here, we describe the construction of an ordered E. coli CRISPRi collection, designed to knock down the expression of 356 essential genes with the induction of a catalytically inactive Cas9, harbored on the conjugative plasmid pFD152. This mobile CRISPRi library can be conjugated into other ordered genetic libraries to assess combined effects of essential gene knockdowns with non-essential gene deletions. As proof of concept, we probed cell envelope synthesis with two complementary crosses: (1) an Lpp deletion into every CRISPRi knockdown strain and (2) the lolA knockdown plasmid into the Keio collection. These experiments revealed a number of notable genetic interactions for the essential phenotype probed and, in particular, showed suppressing interactions for the loci in question.

This article provides an overview of microbial host selection, synthetic biology, genome annotation, metabolic modeling, and computational methods for predicting gene essentiality for developing a microbial chassis. This article focuses on lactic acid bacteria (LAB) as a microbial chassis and strategies for genome annotation of the LAB genome. As a case study, Lactococcus lactis is chosen based on its well-established therapeutic applications such as probiotics and oral vaccine development. In this article, we have delineated the strategies for genome annotations of lactic acid bacteria. These strategies also provide insights into streamlining genome reduction without compromising the functionality of the chassis and the potential for minimal genome chassis development. These insights underscore the potential for the development of efficient and sustainable synthetic biology systems using streamlined microbial chassis with minimal genomes.

Candida auris is an emerging human pathogen, associated with antifungal drug resistance and hospital candidiasis outbreaks. In this work, we present iRV973, the first reconstructed Genome-scale metabolic model (GSMM) for C. auris. The model was manually curated and experimentally validated, being able to accurately predict the specific growth rate of C. auris and the utilization of several sole carbon and nitrogen sources. The model was compared to GSMMs available for other pathogenic Candida species and exploited as a platform for cross-species comparison, aiming the analysis of their metabolic features and the identification of potential new antifungal targets common to the most prevalent pathogenic Candida species. From a metabolic point of view, we were able to identify unique enzymes in C. auris in comparison with other Candida species, which may represent unique metabolic features. Additionally, 50 enzymes were identified as potential drug targets, given their essentiality in conditions mimicking human serum, common to all four different Candida models analysed. These enzymes represent interesting drug targets for antifungal therapy, including some known targets of antifungal agents used in clinical practice, but also new potential drug targets without any human homolog or drug association in Candida species.

Microbial genomics studies focusing on the dynamics of selection have often used a small number of distant genomes. As a result, they could only analyze mutations that had become fixed during the divergence between species. However, thousands of genomes of some species are now available in public databases, thanks to high-throughput sequencing. These data provide a more complete picture of the polymorphisms segregating within a species, offering a unique insight into the processes that shape the recent evolution of a species. In this study, we present GLASS (Gene-Level Amino-acid Score Shift), a selection test that is based on the predicted effects of amino acid changes. By comparing the distribution of effects of mutations observed in a gene to the expectation in the absence of selection, GLASS can quantify the intensity of selection. We applied GLASS to a dataset of 60,472 Escherichia coli strains and used this to reexamine the longstanding debate about the role of essentiality versus expression level in the rate of protein evolution. We found that selection has contrasting short-term and long-term dynamics, with essential genes being subject to strong purifying selection in the short term, while expression level determines the rate of gene evolution in the long term. GLASS also found an overrepresentation of inactivating mutations in specific transcription factors, such as efflux pump repressors, which is consistent with selection for antibiotic resistance. These gene-inactivating polymorphisms do not reach fixation, suggesting another contrast between short-term fitness gains and long-term counterselection.