Biology subjects

Lim, C. S.

Publications and source records attributed to Lim, C. S..

18 recordsLinked to original sources

bioRxiv2026

Structural basis of CD28 and CTLA-4 interactions with CD80, CD86, and the CD80-PD-L1 heterodimer on artificial and cellular membranes

The opposing actions of the co-stimulatory receptor CD28 and the co-inhibitory receptor CTLA-4, mediated by their interactions with B7-family ligands, govern the balance between T-cell activation and immune tolerance. We determined the cryo-EM structures of CD28 and CTLA-4 bound to CD80, CD86, and the CD80-PD-L1 heterodimer under two-dimensional membrane confinement. We show that CD28-CD80, CD28-CD86, and CTLA-4-CD86 form discrete complexes with closed-leg, cross-leg, and open-leg configurations, respectively, whereas CTLA-4 assembles into extended linear clusters and two-dimensional lattices with CD80. PD-L1 binding to CD80 disrupts CTLA-4-CD80 clustering by preventing CD80 homodimerization, while preserving CD28-CD80 binding through a pronounced architectural rearrangement. We further confirm the formation of linear and two-dimensional CTLA-4-CD80 clusters in cell-derived membrane fragments using cryo-EM. Together, these findings demonstrate that immune checkpoint signaling is governed not only by receptor-ligand affinity but also by membrane-imposed geometry and the competitive reorganization of higher-order assemblies at the immunological synapse.

immunology↗

bioRxiv2026

Iliac vein morphology and wall shear stress: a statistical shape modelling and CFD analysis of patient-specific geometries

Deep vein thrombosis (DVT) is a prevalent vascular condition in which venous anatomy and flow disturbances contribute to the risk of thrombosis, but the mechanistic links between vessel shape and haemodynamics remain poorly quantified. Although computational fluid dynamics (CFD) can estimate flow-related risk metrics such as low wall shear stress (WSS), the influence of anatomical fidelity on these predictions is not well understood. Statistical shape modelling (SSM) offers a principled framework for characterising geometric variability, but its integration with CFD in venous applications is still emerging. This study investigates how different levels of anatomical representation--2D projections, simplified 3D extrusions, and full 3D reconstructions of the common iliac veins--influence both the statistical structure of venous shape variability and the haemodynamic metrics derived from CFD. Using patient-specific MRI/CT data from twelve cases, we constructed SSMs in Deformetrica and performed steady-state CFD simulations in ANSYS Fluent under standardised inflow conditions. We compared the variance structure of the 2D and 3D latent spaces and quantified correlations between principal shape modes and low-WSS burden across three thresholds ([≤] 0.05, 0.10, 0.15 [Pa]). Idealised 3D geometries consistently produced larger low-WSS areas than patient-specific shapes, with average increases of 118-136% across thresholds. The 2D SSM exhibited a strongly hierarchical variance spectrum with one dominant mode that correlated significantly with WSS, whereas the 3D SSM showed a flatter spectrum with weaker univariate associations. These findings demonstrate that geometric fidelity and alignment strategy critically influence shape-flow relationships, highlighting the need for careful model selection when using CFD-based haemodynamic indicators in DVT research. Author summaryDeep vein thrombosis (DVT) is a common condition in which blood clots form in the deep veins of the leg and can lead to serious long-term complications. Although medical imaging captures important anatomical differences between patients, it remains unclear how these variations in vein shape influence local blood flow and the associated risk of clot formation. To address this challenge, we developed a computational framework that combines statistical shape modelling (SSM) with computational fluid dynamics (CFD) to analyse the relationship between venous geometry and haemodynamic risk factors. We examined the common iliac veins at three levels of anatomical detail: simplified two-dimensional projections, intermediate three-dimensional extrusions, and full three-dimensional reconstructions derived from MRI/CT data. By comparing these representations, we show that geometric fidelity strongly affects both the detected modes of anatomical variation and the resulting flow predictions. Simplified geometries consistently overestimated regions of low wall shear stress, a flow feature associated with thrombosis, compared to full 3D models. We also found that shape-flow associations depend heavily on how shapes are aligned and represented. Our findings highlight the importance of anatomical detail in computational venous modelling and provide a foundation for more personalised, simulation-based tools to support DVT treatment.

bioengineering↗

bioRxiv2025

Differential analysis of translation efficiency and usage of open reading frames using DOTSeq

Protein synthesis is regulated by multiple cis-regulatory elements, including small ORFs, yet current differential translation methods assume uniform changes at the gene level. We present DOTSeq, a Differential ORF Translation statistical framework that resolves ORF-level regulation in bulk ribosome profiling (Ribo-seq) experiments and provides ORF-level read summarisation for single-cell Ribo-seq. DOTSeqs core module, Differential ORF Usage (DOU), quantifies changes in an ORFs relative contribution to a genes translation output, using a beta-binomial GLM with flexible dispersion modelling. DOTSeq also implements ORF-level Differential Translation Efficiency (DTE) using a standard approach to complement DOU. Benchmarks show that DOU achieves superior sensitivity with near-nominal FDR across effect sizes, while DTE and some existing methods excel when technical noise is low. DOTSeq introduces an ORF-aware, quantitative framework for ribosome profiling, delivering end-to-end workflows for ORF annotation, read summarisation, contrast estimation, and visualisation to uncover translational control events at scale.

bioinformatics↗

bioRxiv2025

Decoding the interconnected splicing patterns of hepatitis B virus and host using large language and deep learning models

Hepatitis B virus (HBV) infection causes approximately one million deaths annually and remains a major driver of hepatocellular carcinoma. Despite its compact 3.2-kb genome, HBV exhibits extensive alternative splicing. Functionally, HBV splice variants contribute to immune evasion and reduce the likelihood of achieving a functional cure. Here, we show that HBV splicing efficiency--quantified from 279 RNA-seq libraries of HBV-associated liver biopsies and cultured cells--correlates more strongly with disease progression than the overall proportion of spliced HBV RNAs, an emerging biomarker. All HBV splice sites are embedded within protein-coding regions, forming a genetic architecture distinct from typical host splice sites. To decode the sequence determinants of HBV splicing, we apply transformer-based and deep learning models to 4,707 HBV genomes. These models reveal that more highly used HBV splice sites are more conserved and share features with host splice sites that are less frequently used but still functional. This similarity likely reflects constraints imposed by HBVs compact genome, which must accommodate overlapping protein-coding regions. HBV may have evolved to exploit suboptimal but spliceable host-like motifs without disrupting its genetic architecture. Further analysis of splicing propensity across HBV genomes reveals genotype-specific patterns, indicating regulation by sequence context in a site- and genotype-dependent manner. HBV genotypes may have coevolved with their human hosts to fine-tune splicing through host-like features, supporting mechanisms of viral persistence and immune evasion. This study demonstrates the utility of AI in decoding viral splicing architectures and provides a framework for investigating co-transcriptional processes in other clinically important viruses. Impact StatementHepatitis B virus (HBV) is a major global health concern, causing 1.1 million deaths in 2022 and 1.2 million new infections each year. It is a leading cause of serious liver conditions, including cirrhosis and liver cancer. Although vaccines can prevent HBV infection, there is currently no cure. HBV produces different types of genetic messages (RNAs), including spliced versions that are processed by the host cells machinery. These spliced RNAs help the virus evade the immune system and make it harder for treatments to fully clear the infection. In this study, we analysed 279 HBV samples from liver tissues and lab-grown cells and found that the efficiency of host-mediated splicing of viral RNA reflects the severity of disease. Using advanced artificial intelligence tools, we mapped the splicing patterns in both the virus and human host, and investigated over 4,700 HBV genomes. We discovered that HBV splice sites resemble host splice sites that are used less frequently but remain functional, suggesting the virus has evolved RNA sequences that are compatible with the host cells splicing machinery while accommodating its compact genome. Insights into this viral adaptation may help researchers identify new biomarkers for disease severity and develop therapeutic strategies that disrupt the viruss ability to exploit the host cells machinery. Data SummaryThe raw RNA-sequencing (RNA-seq) libraries analysed in this study were previously published and are available in the Gene Expression Omnibus under accession number GSE155983. These libraries form part of a curated collection of 279 RNA-seq datasets derived from HBV-associated liver biopsy tissues and cultured cells. Further details and metadata are available in the associated GitHub repository: https://github.com/lcscs12345/HBV_splicing_paper_2025.

genomics↗

bioRxiv2024

RIBOSS detects novel translational events by combining long- and short-read transcriptome and translatome profiling

Ribosome profiling is a high-throughput sequencing technique that captures the positions of translating ribosomes on RNAs. Recent advancements in ribosome profiling include achieving highly-phased ribosome footprints for plant translatomes and more recently for bacterial translatomes. This substantially increases the specificity of detecting open reading frames (ORFs) that can be translated, such as small ORFs located upstream and downstream of the annotated ORFs. However, most genomes (e.g. bacterial genomes) lack the annotations for the transcription start and termination sites. This hinders the systematic discovery of novel ORFs in the untranslated regions in ribosome profiling data. Here we develop a new computational pipeline called RIBOSS. We use RIBOSS to leverage long-read and short-read data for de novo transcriptome assembly, and highly-phased ribosome profiling data for detecting novel translational events in the newly assembled transcriptome. We demonstrate the capability of RIBOSS using recently published metatranscriptome and translatome data for Salmonella enterica serovar Typhimurium. The RIBOSS Python modules are versatile and can be used to analyse prokaryotic or eukaryotic data. In sum, RIBOSS is the first computational pipeline to integrate long- and short-read sequencing technologies to investigate translation. RIBOSS is freely available at https://github.com/lcscs12345/riboss.

bioinformatics↗

bioRxiv2024

Highly ordered clustering of TNFα and BAFF ligand-receptor-adaptor complexes bound to lipid membranes

We report the cryo-EM structures of clusters of TNF receptor family proteins, TNFR1 and BAFFR. The receptor-ligand complexes were anchored to a flat lipid layer to mimic the membrane-bound state. We observed that the TNF-TNFR1 complex forms highly ordered binary, bent, trigonal, and linear quadruple clusters of trimers on the lipid membrane. A non-competitive antagonist of TNFR1 disrupted these clusters without interfering with ligand binding. Moreover, we found that the BAFF-BAFFR complex forms pentagonal, double-pentagonal, or half-spherical clusters of trimers. Mutations in BAFF that inhibit BAFFR receptor activation prevented ordered clustering without disrupting receptor binding. TRAF3 induced a structural shift in the BAFF-BAFFR cluster, resulting in a flat hexagonal cluster. Our data demonstrated that precisely structured clustering is essential for the activation of these receptors. The lipid monolayer method will aid in studying the clusters of other transmembrane proteins and facilitate the discovery of therapeutic agents that regulate their clustering.

biochemistry↗

bioRxiv2023

Computational Modeling of Stapled Coiled-Coil Inhibitors Against Bcr-Abl: Toward a Treatment Strategy for CML

The chimeric oncoprotein Bcr-Abl is the causative agent of virtually all chronic myeloid leukemias (CML) and a subset of acute lymphoblastic leukemias (ALL). As a result of the so-called Philadelphia Chromosome translocation t(9;22), Bcr-Abl manifests as a constitutively active tyrosine kinase which promotes leukemogenesis by activation of cell cycle signaling pathways. Constitutive and oncogenic activation is mediated by an N-terminal coiled-coil oligomerization domain in Bcr (Bcr-CC), presenting a therapeutic target for inhibition of Bcr-Abl activity toward the treatment of Bcr-Abl+ leukemias. Previously, we demonstrated that a rationally designed Bcr-CC mutant, CCmut3, exerts a dominant negative effect upon Bcr-Abl activity by preferential oligomerization with Bcr-CC. Moreover, we have shown conjugation to a leukemia-specific cell-penetrating peptide (CPP-CCmut3) improves intracellular delivery and activity. However, our full-length CPP-CCmut3 construct (81 aa) is encumbered by an intrinsically high degree of conformational variability and susceptibility to proteolytic degradation, relative to traditional small molecule therapeutics. Here, we iterate a new generation of our inhibitor against Bcr-CC mediated Bcr-Abl assembly that is designed to address these constraints through incorporation of all-hydrocarbon staples spanning i, i + 7 positions in helix 2 (CPP-CCmut3-st). We utilize computational modeling and biomolecular simulation to design and characterize single and double staple candidates in silico, evaluating binding energetics and building upon our seminal work modeling single hydrocarbon staples when applied to a truncated Bcr-CC sequence. This strategy enables us to efficiently build, characterize, and screen lead single/double stapled CPP-CCmut3-st candidates for experimental studies and validation in vitro and in vivo. In addition to full-length CPP-CCmut, we model a truncated system characterized by deletion of helix 1 and the flexible-loop linker, which are known to impart high conformational variability. To study the impact of the N-terminal cyclic CPP toward model stability and inhibitor activity, we also model the full-length and truncated systems without CPP, with cyclized CPP, and with linear CPP, for a total of six systems which comprise our library. From this library, we present lead stapled peptide candidates to be synthesized and evaluated experimentally as our next-generation inhibitors against Bcr-Abl.

biophysics↗

bioRxiv2023

p53-Bad* in a Hepatocellular Carcinoma Mouse Model

Recent advances in liver cancer treatments have not changed the fact that the majority of patients will not survive the disease. In order to advance future liver cancer treatments, this work presents an exploration of various iterations of the liver cancer specific AFP promoter as well as the gene construct p53-Bad*. p53-Bad* is a mitochondrially targeted re-engineered p53 therapy that has shown previous success in a zebrafish HCC model. Both the most promising AFP promoter and p53-Bad* were packaged in an adenoviral delivery system and tested in vitro in liver cancer cell lines. Finally, mixed results for adenoviral p53-Bad* in vivo are presented, and this work suggests future modifications to study parameters in order to further explore the potential of p53-Bad* as a potential liver cancer therapeutic.

molecular biology↗

bioRxiv2022

Phylogenetic analysis of human parainfluenza type 3 virus strains responsible for the outbreak during the COVID-19 pandemic in Seoul, South Korea

BackgroundHuman parainfluenza virus 3 (HPIV3) is a major respiratory pathogen that causes acute respiratory infections in infants and children. Since September 2021, an out-of-season HPIV3 rebound has been noted in Korea. The objective of this study was to analyze the molecular characteristics of the HPIV3 strains responsible for the outbreak in Seoul, South Korea. MethodsA total of 61 HPIV3-positive nasopharyngeal swab specimens were collected between October and November 2021. Using 33 HPIV3-positive specimens, partial nucleotide sequences of the HPIV3 hemagglutinin-neuraminidase (HN) gene were aligned with previously published HN gene sequences for phylogenetic and genetic distance (p-distance) analyses. ResultsPhylogenetic tree revealed that all Seoul HPIV3 strains grouped within the phylogenetic subcluster C3. However, these strains formed a unique cluster that branched separately from the C3a lineage. This cluster showed 99% bootstrap support with a p-distance < 0.001. Genetic distances within the other C3 lineages ranged from 0.013 (C3a) to 0.023 (C3c). Deduced amino acid sequences of the HN gene revealed four protein substitutions in Seoul HPIV3 strains that have rarely been observed in other reference strains: A22T, K31N, G387S, and E514K. ConclusionsPhylogenetic analysis of Seoul HPIV3 strains revealed that the strain belonged to a separate cluster within subcluster C3. Genetic distances among strains within subcluster C3 suggest the emergence of a new genetic lineage. The emergence of a new genetic lineage could pose a potential risk of a new epidemic. Further monitoring of the circulating HPIV3 strains is needed to understand the importance of newly discovered mutations.

microbiology↗

bioRxiv2022

LazyPair: scalable prediction of protein-protein interactions and interaction types

MotivationAlmost all cellular processes require protein-protein interactions. Common interaction types include binding, post-translational modifications, and catalysis. However, existing prediction tools do not take these interaction types into account and do not scale well on proteome-wide prediction. ResultsHere we show that a random forest classifier trained on per-residue physicochemical and biochemical properties is useful for predicting protein-protein interactions. Counterintuitively, we find that training random forests by individual interaction types improves accuracy. Furthermore, a combination of these specialised classifiers improves generalisability. We call our protein-protein interaction prediction tool LazyPair. More importantly, LazyPair outperforms the state-of-the-art in accuracy, generalisability and scalability. Availability and implementationLazyPair and the source code and data for reproducing our analysis are freely available at https://github.com/Gardner-BinfLab/PPI_Analysis_2022 and https://doi.org/10.5281/zenodo.6071630. The web server version and the source code are freely available at https://tisigner.com/lazypair/ and https://github.com/Gardner-BinfLab/TISIGNER-ReactJS, respectively.

bioinformatics↗

bioRxiv2021

Fast off-rate CD229 chimeric antigen receptor T cells efficiently target multiple myeloma, spare T cells, and exhibit reduced trogocytosis

T cells expressing chimeric antigen receptors have shown remarkable therapeutic activity against different types of cancer. However, their wider use has been hampered by the potential for life-threatening toxicities due to the unintended targeting of healthy cells expressing low levels of the targeted antigen. We have now developed an affinity-tuning approach for the generation of minimally modified, low-affinity antibody variants derived from existing high-affinity antibodies. Using this approach, we engineered low affinity variants of the fully human CD229-specific antibody 2D3. Parental 2D3 originally efficiently targeted multiple myeloma cells but also healthy T cells expressing low levels of CD229. We demonstrate that CAR T cells based on a low affinity variant of 2D3, engineered to also express CJUN to increase CAR T cell expansion, maintain the parental antibodys anti-tumor activity but lack its targeting of healthy T cells in vitro and in vivo. In addition, we found that low affinity CD229 CAR T cells show reduced trogocytosis potentially augmenting CAR T cell persistence. The fast off-rate CAR produced using our affinity tuning approach eliminates a key liability of CD229 CAR T cells and paves the way for the effective and safe treatment of patients with multiple myeloma and other lymphoid malignancies. One sentence summaryRational T cell engineering yields low affinity CD229 CAR T cells overexpressing CJUN, which maintain the parental cells anti-tumor activity but eliminate killing of healthy T cells, increasing CAR T cell expansion, and decreasing trogocytosis.

bioengineering↗

bioRxiv2021

Effective Computational Framework for Pre-Interventional Planning of Peripheral Arteriovenous Malformations with In Vivo and In Vitro Validation

PurposePeripheral arteriovenous malformations (pAVMs) are congenital lesions characterised by abnormal high-flow, low-resistance vascular connections - constituting the so-called nidus - between arteries and veins. The mainstay treatment typically involves the embolisation of the nidus with embolic and sclerosant agents, however the complexity of AVMs often leads to uncertain outcomes. This study aims at developing a simple, yet effective computational framework to aid the clinical decision making around the treatment of pAVMs. MethodsA computational model was developed to simulate the pre-, intra-, and post-intervention haemodynamics of an AVM. A porous medium of varying permeability was used to simulate the effect that the sclerosant has on the blood flow through the nidus. The computational model was informed by computed tomography (CT) scans and digital subtraction angiography (DSA) images, and the results were compared against clinical data and experimental results. ResultsThe computational model was able to simulate the blood flow through the AVM throughout the intervention and predict (direct and indirect) haemodynamic changes due to the embolisation. The simulated transport of the dye in the AVM was compared against DSA time-series obtained at different intervention stages, providing confidence in the results. Moreover, experimental data obtained via a mock circulatory system involving a patient specific 3D printed phantom of the same AVM provided further validation of the simulation results. ConclusionWe developed a simple computational framework to simulate AVM haemodynamics and predict the effects of the embolisation procedure. The developed model lays the foundation of a new, computationally driven treatment planning tool for AVM embolisation procedures.

bioengineering↗

bioRxiv2020

Annotating eukaryotic and toxin-specific signal peptides using Razor

MotivationSignal peptides are responsible for protein transport and secretion and are ubiquitous to all forms of life. The annotation of signal peptides is important for understanding protein translocation and toxin secretion and evolution. ResultsHere we explore the features of these signal sequences from eukaryotic proteins. Strikingly, we find that the signal peptides from secretory toxins have common features across kingdoms, supporting the idea of horizontal gene transfer or convergence of toxin genes across kingdoms. We leverage these features to build Razor, a simple yet powerful tool specialised in identifying signal peptides from toxins using the first 23 N-terminal residues. We demonstrate the usability of Razor by analysing all the sequences reviewed by UniProt. Indeed, Razor is able to identify toxins using their N-terminal sequences only. Strikingly, we also discover that many defensive proteins across kingdoms harbour a toxin-like signal peptide; some of these defensive proteins have emerged through convergent evolution, e.g. defensin and defensin-like protein families, and phospholipase families. In sum, Razor uses an approach independent of homology search to identify novel and known toxin classes across species using N-terminal residues. Availability and implementationRazor is available as a web application (https://tisigner.com/razor) and a command-line tool (https://github.com/Gardner-BinfLab/Razor).

bioinformatics↗

bioRxiv2020

Quantitative analysis of the splice variants expressed by the major hepatitis B virus genotypes

Hepatitis B virus (HBV) is a major human pathogen that causes liver diseases. The main HBV RNAs are unspliced transcripts that encode the key viral proteins. Recent studies show that some of the HBV spliced transcript isoforms are predictive of liver cancer, yet the roles of these spliced transcripts remain elusive. Furthermore, a total of 9 major HBV genotypes were isolated from discrete geographical regions of the world, it is likely that these genotypes may express a broad variety of spliced transcript isoforms. To systematically study the HBV splice variants, we transfected the human hepatoma cells Huh7 with 4 HBV genotypes (A2, B2, C2, and D3), followed by deep RNA-sequencing. We found that 12-25% of HBV RNAs were splice variants, which were reproducibly detected across independent biological replicates. This accounted for a total of 6 novel and 6 previously identified splice variants. In particular, 2 highly abundant novel splice variants, in which we called the putative splice variants 1 and 5 (pSP1 and pSP5), were specifically expressed at high levels in genotypes D3 and B2, respectively. In general, the HBV splicing profiles varied across the genotypes except for the known spliced pgRNAs SP1 and SP9, which were present in all 4 major genotypes. Counterintuitively, these singly spliced SP1 and SP9 had a suboptimal 5' splice site, suggesting that splicing of HBV RNAs is tightly controlled by the viral post-transcriptional regulatory RNA element. IMPORTANCEHBV infection affects over 257 million people worldwide. HBV is a major cause of liver diseases including cancer and there is no cure. Some HBV RNAs are spliced variants and their roles are largely unclear, although some splice variants have been previously found to be associated with liver cancer. HBV exists as 9 genotypes worldwide with marked differences in replicative capacity and disease sequelae. Whether HBV splice variants vary for the different genotypes is yet to be investigated in depth. Here we sequenced RNAs from 4 major HBV genotypes using a cell culture system. We found 6 new and 6 previously known splice variants across these genotypes. Some novel splice variants were present at high levels, suggesting they could be functionally important. Interestingly, although HBV has adapted to human hosts for over 50,000 years, the most frequently spliced location shared little flanking sequence similarity with that of humans.

microbiology↗

bioRxiv2020

Analysis of fungal genomes reveals commonalities of intron loss/gain and functions in intron-poor species

Current evolutionary reconstructions predict that early eukaryotic ancestors including both the last common ancestor of eukaryotes and of all fungi had intron-rich genomes. However, some extant eukaryotes have few introns, raising the question as to why these few introns are retained. Here we have used recently available fungal genomes to address this question. Evolutionary reconstruction of intron presence and absence using 263 diverse fungal species support the idea that massive intron loss has occurred in multiple clades. The intron densities estimated in the fungal ancestral states differ from zero to 8.28 introns per one kbp of protein-coding gene. Massive intron loss has occurred not only in microsporidian parasites and saccharomycetous yeasts (0.01 and 0.05 introns/kbp on average, respectively), but also in diverse smuts and allies (e.g. Ustilago maydis, Meira miltonrushii and Malassezia globosa have 0.06, 0.10 and 0.20 introns/kbp, respectively). To investigate the roles of introns, we searched for their special characteristics using 1302 orthologous genes from eight intron-poor fungi. Notably, most of these introns are found close to the translation initiation codons. Our transcriptome and translatome data analyses showed that these introns are from genes with both higher mRNA expression and translation efficiency. Furthermore, these introns are common in specific classes of genes (e.g. genes involved in translation and Golgi vesicle transport), and rare in others (e.g. base-excision repair genes). Our study shows that fungal introns have a complex evolutionary history and underappreciated roles in gene expression.

evolutionary biology↗

bioRxiv2020

Protein solubility is controlled by global structural flexibility

MotivationRecombinant protein production is a widely used technique in the biotechnology and biomedical industries, yet only a quarter of target proteins are soluble and can therefore be purified. ResultsWe have discovered that global structural flexibility, which can be modeled by normalised B-factors, accurately predicts the solubility of 12,216 recombinant proteins expressed in Escherichia coli. We have optimised B-factors, and derived a new set of values for solubility scoring that further improves prediction accuracy. We call this new predictor the Solubility-Weighted Index (SWI). Importantly, SWI outperforms many existing protein solubility prediction tools. Furthermore, we have developed SoDoPE (Soluble Domain for Protein Expression), a web interface that allows users to choose a protein region of interest for predicting and maximising both protein expression and solubility. AvailabilityThe SoDoPE web server and source code are freely available at https://tisigner.com/sodope and https://github.com/Gardner-BinfLab/TISIGNER-ReactJS, respectively. The code and data for reproducing our analysis can be found at https://github.com/Gardner-BinfLab/SoDoPE_paper2020.

bioinformatics↗

bioRxiv2019

Highly accessible translation initiation sites are predictive of successful heterologous protein expression

Recombinant protein production is a key process in generating proteins of interest in the pharmaceutical industry and biomedical research. However, about 50% of recombinant proteins fail to be expressed in a variety of host cells. To address this problem, we modified up to the first nine codons of messenger RNAs with synonymous substitutions and showed that protein levels can be tuned. These modifications alter the accessibility of translation initiation sites. We also reveal the dynamics between accessibility, gene expression, and turnovers using a coarse-grained simulation.

bioinformatics↗

bioRxiv2019

SVCurator: A Crowdsourcing app to visualize evidence of structural variants for the human genome

A high quality benchmark for small variants encompassing 88 to 90% of the reference genome has been developed for seven Genome in a Bottle (GIAB) reference samples. However a reliable benchmark for large indels and structural variants (SVs) is yet to be defined. In this study, we manually curated 1235 SVs which can ultimately be used to evaluate SV callers or train machine learning models. We developed a crowdsourcing app - SVCurator - to help curators manually review large indels and SVs within the human genome, and report their genotype and size accuracy.\n\nSVCurator is a Python Flask-based web platform that displays images from short, long, and linked read sequencing data from the GIAB Ashkenazi Jewish Trio son [NIST RM 8391/HG002], We asked curators to assign labels describing SV type (deletion or insertion), size accuracy, and genotype for 1235 putative insertions and deletions sampled from different size bins between 20 and 892,149 bp. The crowdsourced results were highly concordant with 37 out of the 61 curators having at least 78% concordance with a set of expert curators, where there was 93% concordance amongst expert curators. This produced high confidence labels for 935 events. When compared to the heuristic-based draft benchmark SV callset from GIAB, the SVCurator crowdsourced labels were 94.5% concordant with the benchmark set. We found that curators can successfully evaluate putative SVs when given evidence from multiple sequencing technologies.

genomics↗