Poster numbers will be assigned May 30th.
If you can not find your poster below that probably means you have not yet confirmed you will be attending ISMB/ECCB 2015. To confirm your poster find the poster acceptence email there will be a confirmation link. Click on it and follow the instructions.

If you need further assistance please contact and provide your poster title or submission ID.

Category J - 'Pathogen informatics'
J01 - Bacterial vaccine design using reverse vaccinology
Ashley Heinson, University of Southampton,
Carmen Denman, London School of Hygiene and Tropical Medicine,
Yawwani Gunawardana, University of Southampton,
Mahesan Niranjan, University of Southampton,
Bastiaan Moekser, University of Southampton,
Christopher Woelk, University of Southampton,
Short Abstract: Reverse Vaccinology (RV) uses computational approaches to identify vaccine candidates in the genomes of bacterial pathogens. Vaccine development for bacterial pathogens is at a critical juncture due to widespread antibiotic resistance. Previously our group was the first to apply machine learning approaches to the identification of vaccine candidates in an RV pipeline. The current study aims to dramatically enhance RV by increasing the size of the training data, expanding the number of bioinformatics programs with biological relevance used for protein annotation, and employing nested cross-validation. A literature search identified 200 vaccine candidates, defined as a protein that resulted in significant protection in an animal model following immunization and subsequent challenge with a bacterial pathogen. This positive training data was twinned with negative training data and annotated with 30 bioinformatic tools capable of annotating protein data to derive a total of 200 annotation features. A support vector machine was trained on this data and compared to previous analyses that used smaller training data sets, less protein annotation tools, and improper models of cross-validation. Although nested cross validation led to a reduction in accuracy compared to previous methods (that were over fit), increasing the size of the training data set and expanding the number of protein annotation tools led to higher accuracies (>92%). In conclusion, we have dramatically improved previous RV approaches such that our trained classifier can now be used to select novel vaccine candidates in the genomes of bacterial pathogens for validation in animal models.
J02 - Multiple genomic and transcriptomic switches drive oxacillin resistance in community-acquired MRSA
Simone Coughlan, , Ireland
Justine Rudkin, NUI Galway, Ireland
James P O\'Gara, NUI Galway, Ireland
Tim Downing , Dublin City University , Ireland
Short Abstract: Community acquired methicillin resistant Staphylococcus aureus (CA-MRSA) is resistant to beta-lactam antibiotics and is transmitted among otherwise healthy individuals who have not been recently hospitalised. To discover genetic and transcriptional controls of oxacillin resistance in CA-MRSA, a heterogenously resistant (HeR) USA300 S. aureus isolate was grown in vitro at 100 ug/ml oxacillin so that it became homogenously resistant (HoR). The original HeR isolate and HoR sample were grown with 0, 0.5, and 2 ug/ml oxacillin: their genomes and transcriptomes were sequenced using Illumina Miseq with 300 bp paired-end reads.

Genomic analysis revealed distinct responses to high and low oxacillin doses. At 100 ug/ml, cells gained a stop mutation in gdpP, a c-di-AMP phosphodiesterase involved in the stringent stress response. At low 0.5-2 ug/ml doses, half of the HeR samples had a missense mutation in rpoB, which encodes the beta subunit of DNA-directed RNA polymerase.

This opposing reaction to high and low doses of oxacillin was matched at the RNA level. Once the cells had become resistant to 100 ug/ml, few (n=13) genes were differentially expressed in at least one of three comparisons (0-0.5, 0-2, 0.5-2 ug/ml). By contrast, 248 genes were differentially expressed in at least one of three equivalent comparisons in the HeR samples.

Thus, low doses of oxacillin invoke a response that primarily involves transcriptional changes, whereas resistance to high doses is enabled by DNA mutation. These findings highlight mechanisms employed by CA-MRSA to survive and reproduce in the presence of high concentrations of antibiotics.
J03 - Extending matrix completion for multi-source transfer learning in host-pathogen protein interaction networks
Meghana Kshirsagar, Carnegie Mellon University, United States
Jaime Carbonell, Carnegie Mellon University, United States
Judith Klein-Seetharaman, Systems Biology Centre, University of Warwick,
Short Abstract: Newly emerging viral diseases, such as swine flu, SARS and the recent Ebola epidemic lead to wide-spread loss of life and health. In the absence of a complete molecular-level understanding of how new diseases work, it is vital that we combine knowledge using the available information from other diseases. Towards this goal, we employ multiple-source transfer learning. Each source task comprises the host-pathogen protein interactions corresponding to one disease and our goal is to infer the interactions in a new disease (target task) for which no or very few interactions (training data) are available. We model the protein interaction prediction problem as a matrix completion task, where each entry of the matrix corresponds to a (host, pathogen) protein pair. The observed entries of the matrix represent the known interactions and we wish to predict the remainder. Within this formulation, we present novel transfer learning techniques that exploit the unlabeled data from the target task, in addition to the labeled interactions on the source tasks.

Our transfer learning based methods outperform baselines in quantitative
metrics on held-out target tasks: HIV, Influenza and Hepatitis. We observe that more related viruses are better source tasks. We analyze our multi-source model and find interesting patterns in the learned parameters. We further apply our method on predicting protein interactions between Ebola virus proteins and human proteins where little protein interactions data is available. The top ranked novel predictions involve the Ebola glycoprotein with human proteins involved in ubiquitination, Calcium signalling and with the cytoskeleton.
J04 - Sequence and structural diversity of transferrin receptors in Gram-negative porcine pathogens
Jamie Fegan, University of Calgary, Canada
Dave Curran, University of Calgary, Canada
Paul Adamiak, University of Calgary, Canada
Chenzhe Qian, University of Calgary, Canada
Rong-hua Yu, University of Calgary, Canada
Anthony Schryvers, University of Calgary, Canada
Short Abstract: Actinobacillus pleuropneumoniae, A. suis, and Haemophilus parasuis are clinically important porcine pathogens where immunization is an important strategy for disease prevention. An attractive target for vaccine development is the surface receptor involved in iron acquisition, since it is common to all three species and essential for survival and disease causation. A previous study showed an engineered antigen derived from the lipoprotein component of the receptor, transferrin-binding protein B (TbpB), is effective at preventing infection by H. parasuis. This study was initiated to examine the genetic and immunogenic diversity of the transferrin receptor system from these species to enhance our ability to predict the breadth of protection against them.
To evaluate the sequence diversity, nucleic acid sequences were obtained from a diverse collection of isolates from all three species. Phylogenetic analyses using M-Coffee and Phyml demonstrated that TbpB protein sequences cluster independently of species, indicating that there is extensive exchange between these species such that receptor-based vaccines should target all three.
To evaluate the ability of TbpB-derived antigens to induce a cross-reactive response, pigs were immunized with intact TbpB, TbpB N-lobe or TbpB C-lobe from A. pleuropneumoniae strain H49 and the resulting sera were tested against a panel of TbpBs; demonstrating the C-lobe induces a broadly cross-reactive response.
Overall our results indicate that there is a multi-species reservoir for transferrin receptor antigenic variation amongst these pathogens. While this presents a challenge to future vaccine development, our results also suggest a rationally designed TbpB-based vaccine may provide protection against all three pathogens.
J05 - IRIDA: Canada’s federated platform for genomic epidemiology
Damion Dooley, BC Centre for Disease Control / UBC, Canada
Josh Adam, National Microbiology Laboratory, Public Health Agency of Canada, Canada
Franklin Bristow, National Microbiology Laboratory, Public Health Agency of Canada, Canada
Thomas Matthews, National Microbiology Laboratory, Public Health Agency of Canada, Canada
Aaron Petkau, National Microbiology Laboratory, Public Health Agency of Canada, Canada
João André Carriço, Faculty of Medicine, University of Lisbon, Portugal
Mélanie Courtot, BC Public Health Microbiology and Reference Laboratory; Simon Fraser University, Canada
Peter Kruczkiewicz, Laboratory for Foodborne Zoonoses, Public Health Agency of Canada, Canada
Emma Griffiths, Simon Fraser University, Canada
Matthew Laird, Simon Fraser University, Canada
Judy Isaac-Renton, BC Public Health Microbiology and Reference Laboratory, Canada
Alex Keddy, Dalhousie University, Canada
Lynn Schriml, University of Maryland School of Medicine, United States
Eduardo Taboada, Laboratory for Foodborne Zoonoses, Public Health Agency of Canada, Canada
Patrick Tang, BC Public Health Microbiology and Reference Laboratory; 2University of British Columbia, Canada
Robert G. Beiko, Dalhousie University , Canada
Morag Graham, National Microbiology Laboratory, Public Health Agency of Canada, Canada
Gary Van Domselaar, National Microbiology Laboratory, Public Health Agency of Canada, Canada
William Hsiao, C Public Health Microbiology and Reference Laboratory; 2University of British Columbia, Canada
Fiona Brinkman, 5Simon Fraser University, Canada
Short Abstract: Whole-genome sequencing (WGS) can provide researchers, epidemiologists, and other public health workers with a high-resolution snapshot of bacterial samples collected during an infectious disease outbreak, improving their ability to identify virulence features and show definitive, fine-grained relationships between isolates. However, existing genomic data analysis tools and pipelines remain prohibitively complex and require considerable technical knowledge to implement and maintain. Additionally, sharing genomic data and meta-data securely between health agencies remains challenging. To address these issues, Canada’s open source Integrated Rapid Infectious Disease Analysis (IRIDA) platform will equip public health workers with turnkey easy-to-use genomic data analysis software. This vision is supported by IRIDA’s multi-disciplinary team consisting of 5 working groups: 1) Ontology and Database; 2) Microbial Typing; 3) Architecture and API; 4) Tools Development; 5) User Experience. IRIDA’s web application insulates users from the complexity of data analysis pipelines, which are processed by an internal Galaxy server (, and simplifies reporting of the resulting analysis, provenance and quality control data. To facilitate collaborations among agencies, fine-grained access control lists and a common REST API enable localized stand-alone platform instances to access shared (federated) data. An overall OBO-compliant ( application ontology suite - ranging from lab, clinical, environmental and pathogen sample terms, to food and location epidemiology metadata – guides the import and export of project data. We will present the IRIDA architecture and how the implemented analysis functionalities can be used to facilitate outbreak investigations. More information can be found at
J06 - The Comprehensive Antibiotic Resistance Database
Andrew McArthur, McMaster University, Canada
Nicholas Waglechner, McMaster University, Canada
Fazmin Nizam, McMaster University, Canada
Sheldon Pereira, McMaster University, Canada
Baofeng Jia, McMaster University, Canada
Daim Sardar, McMaster University, Canada
Erin Westman, McMaster University, Canada
Andrew Pawlowski, McMaster University, Canada
Tim Johnson, McMaster University, Canada
Raymond Lo, Simon Fraser University, Canada
Melanie Courtot, Simon Fraser University, Canada
Fiona Brinkman, Simon Fraser University, Canada
Laura Williams, Providence College, United States
Jonathan Frye, U.S. Department of Agriculture, United States
Gerard Wright, McMaster University, Canada
Short Abstract: Antimicrobial resistance (AMR) is among the most pressing public health crises of the 21st Century. Despite the importance of resistance to health, this field has been slow to take advantage of genome scale tools. Phenotype based criteria dominate the epidemiology of antibiotic action and effectiveness. There is a poor understanding of which antibiotic resistance genes are in circulation, which a threat, and how clinicians and public health workers can manage the crisis of resistance. However, DNA sequencing is rapidly decreasing in cost and as such we are on the cusp of an age of high-throughput molecular epidemiology. What are needed are tools for rapid, accurate analysis of DNA sequence data for the genetic underpinnings of antibiotic resistance. In an effort to address this problem, we have created the Comprehensive Antibiotic Resistance Database ( This database is a rigorously curated collection of known antibiotics, targets, and resistance determinants. It integrates disparate molecular and sequence data, provides a unique organizing principle in the form of the Antibiotic Resistance Ontology (ARO), and can quickly identify putative antibiotic resistance genes in raw genome sequences using the novel Resistance Gene Identifier (RGI). Here we review the current state of the CARD, particularly recent advances in the curation of resistance determinants and the structure of the ARO. We will also present our plans for development of semi- and fully-automated text mining algorithms for curation of broader AMR data, construction of Probabilistic Graphic Models for improved AMR phenotype prediction, and development of portable command-line genome analysis tools.
J07 - Comparison of TAL effectors and their predicted host targets across diverse strains of the rice bacterial leaf streak pathogen Xanthomonas oryzae pv. oryzicola
Katherine Wilkins, Cornell University, United States
Nicholas Booher, Cornell University, United States
Li Wang, Cornell University, United States
Adam Bogdanove, Cornell University, United States
Short Abstract: Bacterial leaf streak of rice, caused by Xanthomonas oryzae pv. oryzicola (Xoc), can cause yield losses of 30% in this staple crop. Disease progression is mediated in part by pathogen-secreted transcription activator-like (TAL) effectors that bind host gene promoters to upregulate corresponding genes. Across Xanthomonas species, TAL effectors upregulate susceptibility genes critical for disease and or resistance genes that trigger a host defense response. Knowledge of these important targets informs breeding of resistant rice varieties. Each TAL effector possesses a central repeat region that determines binding specificity, with repeat variation specifying binding site sequence in a one-to-one fashion based on a degenerate “code”. We sequenced the complete genomes of ten diverse strains of Xoc to identify their TAL effector content and used RNA-Seq to compare the transcriptional response of rice inoculated with each strain. Xoc TAL effectors are highly conserved compared to those of the closely related X. oryzae pv. oryzae. Phylogenetic analysis suggests Xoc TAL effectors are primarily vertically transmitted, although the only Xoc TAL effector known to upregulate a susceptibility gene appears to have been horizontally transferred. Prediction of TAL effector targets using the TAL effector-DNA recognition code yields many false positives, with predicted binding sites in every rice gene for each Xoc strain. We filtered these predictions using a machine learning classifier we previously developed and requiring correlation between TAL effector presence and gene upregulation across strains. This filtering results in testable numbers of candidate TAL effector targets of potential importance for rice breeding.
J08 - Dissecting the dynamics of influenza A H5N1 protein sequence diversity
Mohammad Asif Khan, Perdana University, Malaysia
Hadia Syahirah, Perdana University, Malaysia
Tan Swan, Perdana University, Malaysia
Thomas J August, Johns Hopkins University, United States
Short Abstract: Background: The influenza A virus subtype H5N1 is a global concern with potential for a pandemic threat. The sequence diversity exhibited by the virus makes vaccine development a challenging task. Towards this end, there is a need for more detailed and quantitative analysis of the virus mutational changes, including the composition and incidence of the different variants of the viral proteome.

Methodology: Influenza A (H5N1) avian and human protein sequences, both partial and full-length, were retrieved from the publicly available databases. Shannon’s entropy calculations were performed on 9,408 overlapping nonamer positions of the proteome to study the diversity in the context of the immune response. To further dissect the sequence diversity, each aligned overlapping nonamer position was quantified for the incidence (% frequency) of the different sequence motifs, namely index, major variant, minor variants and unique variants.

Results: Notably, the avian proteome was on average more conserved than human. Each motif had a distinctive pattern of incidence in relation to increased total variants. The major variant was a transient form with a maximum incidence of ~46% (avian) and 49% (human). The multiple different minor variants, each with an incidence less than that of the major variant, comprised the principle variant motif, particularly with total variants of ≥ 50%. The unique variants were enigmatic as they were present for nearly all of the positions with total variants < 80%, albeit with a low incidence of 11-14%.

Conclusion. The three variant motifs represent inherent patterns in the organization of the different sequences that function in H5N1 quasispecies fitness-selection.
J09 - Automated whole genome assembly of deep sequenced Hepatitis C viral populations
Bede Constantinides, University of Manchester,
Richard Myers, Public Health England,
Short Abstract: The mutation rates of viruses are unparalleled in nature, and together with the typically short generation times can allow a single founding virus to rapidly establish a genetically diverse population within a host. Accordingly, chronic human infections of RNA viruses such as HIV and hepatitis C virus (HCV) afford plenty of time for viral diversification and adaptation to occur, which may lead to immune escape and drug resistance. Such genetic diversity also poses considerable challenges for sequencing and assembling viral genomes, which often bear little resemblance to available reference sequences. Even where sequencing reads cover the entire target genome, de novo assembly typically gives rise to short, fragmented assemblies which have limited utility in downstream analyses. However, by using in silico normalisation to discard a large proportion of sequencing reads, we show that not only is the computational burden of de novo assembly vastly reduced, but also that complete, contiguous assemblies may be attained. We present a freely available species–agnostic pipeline for assembling whole genome viral sequences, which intelligently evaluates a range of normalisation and de novo assembly parameters in order to yield to an optimal assembly, to which original reads are subsequently aligned for scrutiny. Detailed graphical assembly reports are generated for each input sample, and a batch mode provides an additional end-of-run report comparing individual samples. Finally, we present our application of this pipeline to clinical Hepatitis C samples provided by Public Health England.
J10 - Antigenic variation and recent evolution of the dengue virus envelope protein
Sarah Keasey, USAMRIID, United States
Short Abstract: Dengue is an infectious disease that is now endemic to most tropical or semi-tropical regions of the world. The dengue virus (DENV) circulates in four serotypes (DENV-1, 2, 3, and 4) within the human population, and antibody responses provide incomplete protection across the distinct serotypes. The DENV envelope protein is a primary target of neutralizing antibodies, suggesting that persistence of autochthonous infection cycles is partially driven by selective evolution of this receptor molecule. We examined antigenic surfaces of the DENV-2 envelope protein from isolates collected over seven decades. Bayesian analysis of envelope protein sequences estimated the emergence of the DENV-2 serotype in the late 1800s and revealed significant changes in amino acid residues that were linked to progressive time of isolation. Protein phylogenies recapitulated known aspects of DENV genome evolution, for example divergence from sylvatic strains and clustering of sequences by geographical location of virus isolation from human infections. Using entropy as a measure of sequence variability, we identified highly substituted residues that mapped to surface-exposed regions of the envelope protein monomer as displayed on the mature virus. Notably, a 50% increase in the number of variable surface residues is evident in the envelope protein of DENV-2 isolates in comparison to the other serotypes, and these highly-variable surfaces are dominated by residues that are targets of neutralizing antibodies. The discordant evolution of the DENV-2 envelope protein compared to other viral serotypes should be considered in efforts to control infection by vaccination.
J11 - Immunoinformatics and disorder prediction: an integrative approach for vaccine and diagnostic targets discovery
Daniela Resende, Centro de Pesquisas René Rachou, Brazil
Alexandre Barbosa Reis, Universidade Federal de Ouro Preto, Brazil
Jeronimo Conceição Ruiz, Centro de Pesquisas René Rachou/Fiocruz Minas, Brazil
Short Abstract: Immunoinformatics is an innovative strategy for selection of targets for vaccine and diagnostics with reduced time and costs. Data mining of essential sequences for eliciting protective immune responses through immunoinformatics has been used for indicating good vaccine candidates for Neisseria meningitides and Staphylococcus aureus showing the efficacy of this approach. It was also shown that instrinsically disorded proteins play important role in trypanosomatids virulence. Our hypothesis is that protein disordered regions could be related to immunogenic epitopes facilitating their exposure to the immune system. In this work, we developed a computational approach that integrates: a) T and B cell epitope predictors, namely: NetCTL and NetMHC for T CD8+ epitope prediction; NetMHCII for T CD4+ epitope prediction; and BepiPred for B cell epitope prediction; and b) structural disorder predictors, namely: DisEMBL, IUPred, GlobPipe and VSL2B. In addition, data associated with subcellular location predictions performed by the algorithms WoLF PSORT (eucariotic genomes), PSORTb (procariotic genomes), Sigcleave (signal peptides) and TMHMM (transmembrane domains) were integrated in a relational database.
The workflow had been used for searching vaccine or diagnostic targets in procariotic and eucariotic organisms, including Leishmania infantum, Leishmania braziliensis, Schistosoma mansoni and Ehrlichia canis. Experiments in wet lab are being performed in order to confirm the immunogenicity of the selected proteins from Leishmania and S. mansoni. The correlation between structural disorder and the epitope location will be presented together with the analytical approach developed.
Supported by: CAPES, CNPq, FAPEMIG, UFOP and CPqRR.
J12 - Pathogen identification by the 23S rRNA
Allan Cézar Azevedo Martins, Universidade Federal do Rio de Janeiro, Brazil
Victor Hugo Giordano Dias, Instituto Nacional de Metrologia, Qualidade e Tecnologia, Brazil
Rodrigo Soares Moura-Neto, Universidade Federal do Rio de Janeiro, Brazil
Rosane Silva, Universidade Federal do Rio de Janeiro, Brazil
Short Abstract: Background: The Bioterrorism consists in the intentional and planned utilization of pathogenic agents and toxins to attack a population, in order to promote generalized panic, epidemic outbreaks and local health system overload. These agents are classified in risk groups (A, B or C) accordingly to its outbreak spread and priority. The quickness and the accuracy in pathogens identification to differentiate a Bioterrorist act from a natural epidemic event are essential to an immediate and planned response. Our aim is to investigate 23S rRNA variable regions of biological weapons as a tool for accurate identification of pathogens.
Method: We obtained 19 bacterial 23S rRNA sequences from SILVA database. From the sequences alignment, we generated a consensus sequence for each organism. Thus, we aligned the consensus sequences to identify variables regions.
Result: We identified 37 variable regions and 38 conserved regions. V6, V27, V29, V30 and V34 variable regions present inner conserved regions. Moreover, we clustered different organisms by variable region to obtain an unique variable region that could be used for organism identification.
Conclusion: As previously described in 16S rRNA, 23S rRNA present variable and conserved regions, although the consensus is longest, as expected. It is possible to group differently all the pathogens by variable regions. These groups vary in biodiversity, allowing the identification of the pathogens through the combination of 25 variable regions. The 16S rRNA and specific gene sequences can be used to minimize this number, however, more studies need be done.
Supported by: CAPES, FAPERJ, CNPq and INMETRO.

View Posters By Category

Search Posters: