Giovanni Stracquadanio

Heterologous protein expression is an indispensable strategy for generating recombinant proteins. Escherichia coli (E. coli) is the most widely used microbial host for recombinant protein production due to its rapid growth, well-characterized genetics, and ability to produce recombinant proteins in high yields using modern recombinant DNA technology. However, while there is a plethora of robust protein expression protocols for E. coli, these methods are often unsuitable for high-throughput screening due to their significant resource and time consumption; these protocols are also susceptible to operator error and inconsistency. To address these challenges, we developed Automated Protein EXpression (APEX), a robust and automated pipeline for recombinant protein production in E. coli. APEX leverages the accessible, open-source Opentrons OT-2 platform to automate microbial handling and protein expression with high precision and reproducibility. APEX can be configured to perform heat shock transformation, selective plating, colony sampling, microculturing, and protein expression induction using a low-cost, minimal OT-2 hardware setup. We further demonstrate the efficacy of our automated transformation workflows using a variety of plasmids (2.7-17.7 kb) and exemplify the automated heterologous expression of a diverse array of proteins (27-222 kDa). Designed with customization, modularity, and user-friendliness in mind, APEX can be easily adapted to the operator's needs without requiring any coding expertise. APEX is available at https://github.com/stracquadaniolab/apex-nf under the AGPL3 license.

We present GNN-Suite, a robust modular framework for constructing and benchmarking Graph Neural Network (GNN) architectures in computational biology. GNN-Suite standardises experimentation and reproducibility using the Nextflow workflow to evaluate GNN performance. We demonstrate its utility in identifying cancer-driver genes by constructing molecular networks from protein-protein interaction (PPI) data from STRING and BioGRID and annotating nodes with features from the PCAWG, PID, and COSMIC-CGC repositories. Our design enables fair comparisons among diverse GNN architectures including GAT, GAT3H, GCN, GCN2, GIN, GTN, HGCN, PHGCN, and GraphSAGE and a baseline Logistic Regression (LR) model. All GNNs were configured as standardised two-layer models and trained with uniform hyperparameters (dropout = 0.2; Adam optimiser with learning rate = 0.01; and an adjusted binary cross-entropy loss to address class imbalance) over an 80/20 train-test split for 300 epochs. Each model was evaluated over 10 independent runs with different random seeds to yield statistically robust performance metrics, with balanced accuracy (BACC) as the primary measure. Notably, GCN2 achieved the highest BACC (0.807 +/- 0.035) on a STRING-based network, although all GNN types outperformed the LR baseline, highlighting the advantage of network-based learning over feature-only approaches. Our results show that a common framework for implementing and evaluating GNN architectures aids in identifying not only the best model but also the most effective means of incorporating complementary data. By making GNN-Suite publicly available, we aim to foster reproducible research and promote improved benchmarking standards in computational biology. Future work will explore additional omics datasets and further refine network architectures to enhance predictive accuracy and interpretability in biomedical applications.

The Sc2.0 global consortium to design and construct a synthetic genome based on the Saccharomyces cerevisiae genome commenced in 2006, comprising 16 synthetic chromosomes and a new-to-nature tRNA neochromosome. In this paper we describe assembly and debugging of the 902,994-bp synthetic Saccharomyces cerevisiae chromosome synXVI of the Sc2.0 project. Application of the CRISPR D-BUGS protocol identified defective loci, which were modified to improve sporulation and recover wild-type like growth when grown on glycerol as a sole carbon source when grown at 37˚C. LoxPsym sites inserted downstream of dubious open reading frames impacted the 5’ UTR of genes required for optimal growth and were identified as a systematic cause of defective growth. Based on lessons learned from analysis of Sc2.0 defects and synXVI, an in-silico redesign of the synXVI chromosome was performed, which can be used as a blueprint for future synthetic yeast genome designs. The in-silico redesign of synXVI includes reduced PCR tag frequency, modified chunk and megachunk termini, and adjustments to allocation of loxPsym sites and TAA stop codons to dubious ORFs. This redesign provides a roadmap into applications of Sc2.0 strategies in non-yeast organisms.

Engineering proteins is a challenging task requiring the exploration of a vast design space. Traditionally, this is achieved using Directed Evolution (DE), which is a laborious process. Generative deep learning, instead, can learn biological features of functional proteins from sequence and structural datasets and return novel variants. However, most models do not generate thermodynamically stable proteins, thus leading to many non-functional variants. Here we propose a model called PRotein Engineering by Variational frEe eNergy approximaTion (PREVENT), which generates stable and functional variants by learning the sequence and thermodynamic landscape of a protein. We evaluate PREVENT by designing 40 variants of the conditionally essential E. coli phosphotransferase N-acetyl-L-glutamate kinase (EcNAGK). We find 85% of the variants to be functional, with 55% of them showing similar growth rate compared to the wildtype enzyme, despite harbouring up to 9 mutations. Our results support a new approach that can significantly accelerate protein engineering.

Engineering proteins with desired functions and biochemical properties is pivotal for biotechnology and drug discovery. While computational methods based on evolutionary information are reducing the experimental burden by designing targeted libraries of functional variants, they still have a low success rate when the desired protein has few or very remote homologous sequences. Here we propose an autoregressive model, called Temporal Dirichlet Variational Autoencoder (TDVAE), which exploits the mathematical properties of the Dirichlet distribution and temporal convolution to efficiently learn high-order information from a functionally related, possibly remotely similar, set of sequences. TDVAE is highly accurate in predicting the effects of amino acid mutations, while being significantly 90% smaller than the other state-of-the-art models. We then use TDVAE to design variants of the human alpha galactosidase enzymes as potential treatment for Fabry disease. Our model builds a library of diverse variants which retain sequence, biochemical and structural properties of the wildtype protein, suggesting they could be suitable for enzyme replacement therapy. Taken together, our results show the importance of accurate sequence modelling and the potential of autoregressive models as protein engineering and analysis tools.

Heterologous protein expression is an indispensable strategy for generating recombinant proteins. Escherichia coli ( E. coli ) is the most widely used microbial host for recombinant protein production due to its rapid growth, well-characterised genetics, and ability to produce recombinant proteins in high yields using modern recombinant DNA technology. However, while there is a plethora of robust protein expression protocols for E. coli , these methods are often unsuitable for high-throughput screening due to their significant resource and time consumption; these protocols are also susceptible to operator error and inconsistency. To address these challenges, we have developed Automated Protein EXpression (APEX), a robust and automated protocol for recombinant protein production in E. coli . APEX leverages the accessible, open-source Opentrons OT-2 platform to automate microbial handling and protein expression with high precision and reproducibility. APEX can be configured to perform heat shock transformation, colony selection, colony sampling, microculturing and protein expression induction using a low-cost, minimal OT-2 hardware setup. We further demonstrate the efficacy of our automated transformation workflows using a variety of plasmids (2.7-17.7 kb), and exemplify the automated heterologous expression of a diverse array of proteins (27-222 kDa). Designed with customisation, modularity and user-friendliness in mind, APEX can be easily adapted to the operator’s needs without requiring any coding expertise. APEX is available at https://github.com/stracquadaniolab/apex-nf under the AGPL3 license.

Unlike in bacteria, eukaryotes rarely cluster sets of genes in their genomes according to function, instead having most genes spread randomly across different chromosomes and loci. However, with the advent of genome engineering, synthetic co-location of genes that together encode a cell function has now become possible. Here, using Saccharomyces cerevisiae we demonstrate the feasibility of reorganising a set of yeast genes encoding a cell function, tryptophan biosynthesis, into a synthetic genome module by deleting these genes and their regulatory elements from their native genomic loci while in parallel reconstructing them into gene cluster format by synthetic DNA assembly. As part of synthetic module design, loxPsym sequences recognised by Cre recombinase are placed between all module genes, and we leverage these for a novel master regulation system we call dCreSIR. Using dCreSIR we externally control silencing of synthetic modules by targeted binding of chromatin recruiters to loxPsym sites and this leads to inhibition of local transcription. We further show that dCreSIR can go beyond modules and be used to specifically downregulate expression across an entire synthetic yeast chromosome containing loxPsym sites. Together, our work offers insights into yeast genome organisation and establishes new principles and tools for the future design and construction of modular synthetic yeast genomes.

Extramedullary multiple myeloma (EMM) is an aggressive form of multiple myeloma (MM). This study represents the most comprehensive next-generation sequencing analysis of EMM tumors (N = 14) to date, uncovering key molecular features and describing the tumor microenvironment. We observed the co-occurrence of 1q21 gain/amplification and MAPK pathway mutations in 79% of EMM samples, suggesting that these are crucial mutational events in EMM development. We also demonstrated that patients with mutated KRAS and 1q21 gain/amplification at the time of diagnosis have a significantly higher risk of EMM development (HR = 2.4, p = 0.011) using data from a large CoMMpass dataset. We identified downregulation of CXCR4 and enhanced cell proliferation, along with reduced expression of therapeutic targets (CD38, SLAMF7, GPRC5D, FCRH5), potentially explaining diminished efficacy of immunotherapy. Conversely, we identified significantly upregulated EZH2 and CD70 as potential future therapeutic options. For the first time, we report on the tumor microenvironment of EMM, revealing CD8+ T cells and NK cells as predominant immune effector cells using single-cell sequencing. Finally, this is the first longitudinal study in EMM revealing the molecular changes from the time of diagnosis to EMM relapse.

Biofoundries are automated high-throughput facilities specializing in the design, construction, and testing of engineered/synthetic DNA constructs (plasmids), often from genetic parts. A critical step of this process is assessing the fidelity of the assembled DNA construct to the desired design. Current methods utilized for this purpose are restriction digest or PCR followed by fragment analysis and sequencing. The Edinburgh Genome Foundry (EGF) has recently established a single-molecule sequencing quality control step using the Oxford Nanopore sequencing technology, along with a companion Nextflow pipeline and a Python package, to perform in-depth analysis and generate a detailed report. Our software enables researchers working with plasmids, including biofoundry scientists, to rapidly analyze and interpret sequencing data. In conclusion, we have created a laboratory and software protocol that validates assembled, cloned, or edited plasmids, using Nanopore long-reads, which can serve as a useful resource for the genetics, synthetic biology, and sequencing communities.

Biofoundries are automated high-throughput facilities specialising in the design, construction and testing of engineered/synthetic DNA constructs (plasmids), often from genetic parts. A critical step of this process is assessing the fidelity of the assembled DNA construct to the desired design. Current methods utilised for this purpose are restriction digest or PCR followed by fragment analysis, and sequencing. The Edinburgh Genome Foundry (EGF) has recently established a single-molecule sequencing quality control step using the Oxford Nanopore sequencing technology, along with a companion Nextflow pipeline and a Python package to perform in-depth analysis and generate a detailed report. Our software enables biofoundry scientists and end-users to rapidly analyse sequencing data, without specialised bioinformatics knowledge. In conclusion, we have created a laboratory and software protocol that validates assembled, cloned or edited plasmids, using Nanopore long reads, which can serve as a useful resource for the genetics, synthetic biology and sequencing communities. Author information All authors contributed to the design of the sequencing quality control step and pipeline, and the preparation of the manuscript. P.V. wrote the manuscript, designed and implemented the bioinformatics pipeline and interpreted results. S.D. wrote the manuscript, implemented the laboratory protocol and interpreted results. G.S. designed the bioinformatics pipeline. R.F. wrote the manuscript and contributed to the design of the laboratory protocol and pipeline. Address: Edinburgh Genome Foundry (University of Edinburgh), Michael Swann Building, Max Born Crescent, Edinburgh, EH9 3BF, United Kingdom