Comparative analysis of machine learning algorithms on the microbial strain-specific AMP prediction.

The evolution of drug-resistant pathogenic microbial species is a major global health concern. Naturally occurring, antimicrobial peptides (AMPs) are considered promising candidates to address antibiotic resistance problems. A variety of computational methods have been developed to accurately predict AMPs. The majority of such methods are not microbial strain specific (MSS): they can predict whether a given peptide is active against some microbe, but cannot accurately calculate whether such peptide would be active against a particular MS. Due to insufficient data on most MS, only a few MSS predictive models have been developed so far. To overcome this problem, we developed a novel approach that allows to improve MSS predictive models (MSSPM), based on properties, computed for AMP sequences and characteristics of genomes, computed for target MS. New models can perform predictions of AMPs for MS that do not have data on peptides tested on them. We tested various types of feature engineering as well as different machine learning (ML) algorithms to compare the predictive abilities of resulting models. Among the ML algorithms, Random Forest and AdaBoost performed best. By using genome characteristics as additional features, the performance for all models increased relative to models relying on AMP sequence-based properties only. Our novel MSS AMP predictor is freely accessible as part of DBAASP database resource at http://dbaasp.org/prediction/genome.

DBAASP v3: database of antimicrobial/cytotoxic activity and structure of peptides as a resource for development of new therapeutics.

The Database of Antimicrobial Activity and Structure of Peptides (DBAASP) is an open-access, comprehensive database containing information on amino acid sequences, chemical modifications, 3D structures, bioactivities and toxicities of peptides that possess antimicrobial properties. DBAASP is updated continuously, and at present, version 3.0 (DBAASP v3) contains >15 700 entries (8000 more than the previous version), including >14 500 monomers and nearly 400 homo- and hetero-multimers. Of the monomeric antimicrobial peptides (AMPs), >12 000 are synthetic, about 2700 are ribosomally synthesized, and about 170 are non-ribosomally synthesized. Approximately 3/4 of the entries were added after the initial release of the database in 2014 reflecting the recent sharp increase in interest in AMPs. Despite the increased interest, adoption of peptide antimicrobials in clinical practice is still limited as a consequence of several factors including side effects, problems with bioavailability and high production costs. To assist in developing and optimizing de novo peptides with desired biological activities, DBAASP offers several tools including a sophisticated multifactor analysis of relevant physicochemical properties. Furthermore, DBAASP has implemented a structure modelling pipeline that automates the setup, execution and upload of molecular dynamics (MD) simulations of database peptides. At present, >3200 peptides have been populated with MD trajectories and related analyses that are both viewable within the web browser and available for download. More than 400 DBAASP entries also have links to experimentally determined structures in the Protein Data Bank. DBAASP v3 is freely accessible at http://dbaasp.org.

Feasibility of virtual reality based training for optimising COVID-19 case handling in Uganda.

BACKGROUND: Epidemics and pandemics are causing high morbidity and mortality on a still-evolving scale exemplified by the COVID-19 pandemic. Infection prevention and control (IPC) training for frontline health workers is thus essential. However, classroom or hospital ward-based training portends an infection risk due to the in-person interaction of participants. We explored the use of Virtual Reality (VR) simulations for frontline health worker training since it trains participants without exposing them to infections that would arise from in-person training. It does away with the requirement for expensive personal protective equipment (PPE) that has been in acute shortage and improves learning, retention, and recall. This represents the first attempt in deploying VR-based pedagogy in a Ugandan medical education context. METHODS: We used animated VR-based simulations of bedside and ward-based training scenarios for frontline health workers. The training covered the donning and doffing of PPE, case management of COVID-19 infected individuals, and hand hygiene. It used VR headsets to actualize an immersive experience, via a hybrid of fully-interactive VR and 360° videos. The level of knowledge acquisition between individuals trained using this method was compared to similar cohorts previously trained in a classroom setting. That evaluation was supplemented by a qualitative assessment based on feedback from participants about their experience. RESULTS: The effort resulted in a COVID-19 IPC curriculum adapted into VR, corresponding VR content, and a pioneer cohort of VR trained frontline health workers. The formalized comparison with classroom-trained cohorts showed relatively better outcomes by way of skills acquired, speed of learning, and rates of information retention (P-value = 4.0e-09). In the qualitative assessment, 90% of the participants rated the method as very good, 58.1% strongly agreed that the activities met the course objectives, and 97.7% strongly indicated willingness to refer the course to colleagues. CONCLUSION: VR-based COVID-19 IPC training is feasible, effective and achieves enhanced learning while protecting participants from infections within a pandemic setting in Uganda. It is a delivery medium transferable to the contexts of other highly infectious diseases.

"METAGENOTE: a simplified web platform for metadata annotation of genomic samples and streamlined submission to NCBI's sequence read archive".

BACKGROUND: The improvements in genomics methods coupled with readily accessible high-throughput sequencing have contributed to our understanding of microbial species, metagenomes, infectious diseases and more. To maximize the impact of these genomics studies, it is important that data from biological samples will become publicly available with standardized metadata. The availability of data at public archives provides the hope that greater insights could be obtained through integration with multi-omics data, reproducibility of published studies, or meta-analyses of large diverse datasets. These datasets should include a description of the host, organism, environmental source of the specimen, spatial-temporal information and other relevant metadata, but unfortunately these attributes are often missing and when present, they show inconsistencies in the use of metadata standards and ontologies. RESULTS: METAGENOTE ( https://metagenote.niaid.nih.gov ) is a web portal that greatly facilitates the annotation of samples from genomic studies and streamlines the submission process of sequencing files and metadata to the Sequence Read Archive (SRA) (Leinonen R, et al, Nucleic Acids Res, 39:D19-21, 2011) for public access. This platform offers a wide selection of packages for different types of biological and experimental studies with a special emphasis on the standardization of metadata reporting. These packages follow the guidelines from the MIxS standards developed by the Genomics Standard Consortium (GSC) and adopted by the three partners of the International Nucleotides Sequencing Database Collaboration (INSDC) (Cochrane G, et al, Nucleic Acids Res, 44:D48-50, 2016) - National Center for Biotechnology Information (NCBI), European Bioinformatics Institute (EBI) and the DNA Data Bank of Japan (DDBJ). METAGENOTE then compiles, validates and manages the submission through an easy-to-use web interface minimizing submission errors and eliminating the need for submitting sequencing files via a separate file transfer mechanism. CONCLUSIONS: METAGENOTE is a public resource that focuses on simplifying the annotation and submission process of data with its corresponding metadata. Users of METAGENOTE will benefit from the easy to use annotation interface but most importantly will be encouraged to publish metadata following standards and ontologies that make the public data available for reuse.

A retrospective genomic analysis of drug-resistant strains of M. tuberculosis in a high-burden setting, with an emphasis on comparative diagnostics and reactivation and reinfection status.

BACKGROUND: Recurrence of drug-resistant tuberculosis (DR-TB) after treatment occurs through relapse of the initial infection or reinfection by a new drug-resistant strain. Outbreaks of DR-TB in high burden regions present unique challenges in determining recurrence status for effective disease management and treatment. In the Republic of Moldova the burden of DR-TB is exceptionally high, with many cases presenting as recurrent. METHODS: We performed a retrospective analysis of Mycobacterium tuberculosis from Moldova to better understand the genomic basis of drug resistance and its effect on the determination of recurrence status in a high DR-burden environment. To do this we analyzed genomes from 278 isolates collected from 189 patients, including 87 patients with longitudinal samples. These pathogen genomes were sequenced using Illumina technology, and SNP panels were generated for each sample for use in phylogenetic and network analysis. Discordance between genomic resistance profiles and clinical drug-resistance test results was examined in detail to assess the possibility of mixed infection. RESULTS: There were clusters of multiple patients with 10 or fewer differences among DR-TB samples, which is evidence of person-to-person transmission of DR-TB. Analysis of longitudinally collected isolates revealed that many infections exhibited little change over time, though 35 patients demonstrated reinfection by divergent (number of differences > 10) lineages. Additionally, several same-lineage sample pairs were found to be more divergent than expected for a relapsed infection. Network analysis of the H3/4.2.1 clade found very close relationships among 61 of these samples, making differentiation of reactivation and reinfection difficult. There was discordance between genomic profile and clinical drug sensitivity test results in twelve samples, and four of these had low level (but not statistically significant) variation at DR SNPs suggesting low-level mixed infections. CONCLUSIONS: Whole-genome sequencing provided a detailed view of the genealogical structure of the DR-TB epidemic in Moldova, showing that reinfection may be more prevalent than currently recognized. We also found increased evidence of mixed infection, which could be more robustly characterized with deeper levels of genomic sequencing.

Nephele: a cloud platform for simplified, standardized and reproducible microbiome data analysis.

Motivation: Widespread interest in the study of the microbiome has resulted in data proliferation and the development of powerful computational tools. However, many scientific researchers lack the time, training, or infrastructure to work with large datasets or to install and use command line tools. Results: The National Institute of Allergy and Infectious Diseases (NIAID) has created Nephele, a cloud-based microbiome data analysis platform with standardized pipelines and a simple web interface for transforming raw data into biological insights. Nephele integrates common microbiome analysis tools as well as valuable reference datasets like the healthy human subjects cohort of the Human Microbiome Project (HMP). Nephele is built on the Amazon Web Services cloud, which provides centralized and automated storage and compute capacity, thereby reducing the burden on researchers and their institutions. Availability and implementation: https://nephele.niaid.nih.gov and https://github.com/niaid/Nephele. Contact: darrell.hurt@nih.gov.

Predictive Model of Linear Antimicrobial Peptides Active against Gram-Negative Bacteria.

Antimicrobial peptides (AMPs) have been identified as a potential new class of anti-infectives for drug development. There are a lot of computational methods that try to predict AMPs. Most of them can only predict if a peptide will show any antimicrobial potency, but to the best of our knowledge, there are no tools which can predict antimicrobial potency against particular strains. Here we present a predictive model of linear AMPs being active against particular Gram-negative strains relying on a semi-supervised machine-learning approach with a density-based clustering algorithm. The algorithm can well distinguish peptides active against particular strains from others which may also be active but not against the considered strain. The available AMP prediction tools cannot carry out this task. The prediction tool based on the algorithm suggested herein is available on https://dbaasp.org.

DBAASP v.2: an enhanced database of structure and antimicrobial/cytotoxic activity of natural and synthetic peptides.

Antimicrobial peptides (AMPs) are anti-infectives that may represent a novel and untapped class of biotherapeutics. Increasing interest in AMPs means that new peptides (natural and synthetic) are discovered faster than ever before. We describe herein a new version of the Database of Antimicrobial Activity and Structure of Peptides (DBAASPv.2, which is freely accessible at http://dbaasp.org). This iteration of the database reports chemical structures and empirically-determined activities (MICs, IC50, etc.) against more than 4200 specific target microbes for more than 2000 ribosomal, 80 non-ribosomal and 5700 synthetic peptides. Of these, the vast majority are monomeric, but nearly 200 of these peptides are found as homo- or heterodimers. More than 6100 of the peptides are linear, but about 515 are cyclic and more than 1300 have other intra-chain covalent bonds. More than half of the entries in the database were added after the resource was initially described, which reflects the recent sharp uptick of interest in AMPs. New features of DBAASPv.2 include: (i) user-friendly utilities and reporting functions, (ii) a 'Ranking Search' function to query the database by target species and return a ranked list of peptides with activity against that target and (iii) structural descriptions of the peptides derived from empirical data or calculated by molecular dynamics (MD) simulations. The three-dimensional structural data are critical components for understanding structure-activity relationships and for design of new antimicrobial drugs. We created more than 300 high-throughput MD simulations specifically for inclusion in DBAASP. The resulting structures are described in the database by novel trajectory analysis plots and movies. Another 200+ DBAASP entries have links to the Protein DataBank. All of the structures are easily visualized directly in the web browser.

The TB Portals: an Open-Access, Web-Based Platform for Global Drug-Resistant-Tuberculosis Data Sharing and Analysis.

The TB Portals program is an international consortium of physicians, radiologists, and microbiologists from countries with a heavy burden of drug-resistant tuberculosis working with data scientists and information technology professionals. Together, we have built the TB Portals, a repository of socioeconomic/geographic, clinical, laboratory, radiological, and genomic data from patient cases of drug-resistant tuberculosis backed by shareable, physical samples. Currently, there are 1,299 total cases from five country sites (Azerbaijan, Belarus, Moldova, Georgia, and Romania), 976 (75.1%) of which are multidrug or extensively drug resistant and 38.2%, 51.9%, and 36.3% of which contain X-ray, computed tomography (CT) scan, and genomic data, respectively. The top Mycobacterium tuberculosis lineages represented among collected samples are Beijing, T1, and H3, and single nucleotide polymorphisms (SNPs) that confer resistance to isoniazid, rifampin, ofloxacin, and moxifloxacin occur the most frequently. These data and samples have promoted drug discovery efforts and research into genomics and quantitative image analysis to improve diagnostics while also serving as a valuable resource for researchers and clinical providers. The TB Portals database and associated projects are continually growing, and we invite new partners and collaborations to our initiative. The TB Portals data and their associated analytical and statistical tools are freely available at https://tbportals.niaid.nih.gov/.

Tyrosine sulfation in the second variable loop (V2) of HIV-1 gp120 stabilizes V2-V3 interaction and modulates neutralization sensitivity.

Elicitation of broadly neutralizing antibodies is essential for the development of a protective vaccine against HIV-1. However, the native HIV-1 envelope adopts a protected conformation that conceals highly conserved sites of vulnerability from antibody recognition. Although high-definition structures of the monomeric core of the envelope glycoprotein subunit gp120 and, more recently, of a stabilized soluble gp140 trimer have been solved, fundamental aspects related to the conformation and function of the native envelope remain unresolved. Here, we show that the conserved central region of the second variable loop (V2) of gp120 contains sulfated tyrosines (Tys173 and Tys177) that in the CD4-unbound prefusion state mediate intramolecular interaction between V2 and the conserved base of the third variable loop (V3), functionally mimicking sulfated tyrosines in CCR5 and anti-coreceptor-binding-site antibodies such as 412d. Recombinant gp120 expressed in continuous cell lines displays low constitutive levels of V2 tyrosine sulfation, which can be enhanced markedly by overexpression of the tyrosyl sulfotransferase TPST2. In contrast, virion-associated gp120 produced by primary CD4(+) T cells is inherently highly sulfated. Consistent with a functional role of the V2 sulfotyrosines, enhancement of tyrosine sulfation decreased binding and neutralization of HIV-1 BaL by monomeric soluble CD4, 412d, and anti-V3 antibodies and increased recognition by the trimer-preferring antibodies PG9, PG16, CH01, and PGT145. Conversely, inhibition of tyrosine sulfation increased sensitivity to soluble CD4, 412d, and anti-V3 antibodies and diminished recognition by trimer-preferring antibodies. These results identify the sulfotyrosine-mediated V2-V3 interaction as a critical constraint that stabilizes the native HIV-1 envelope trimer and modulates its sensitivity to neutralization.

Selectivity in the Use of Gi/o Proteins Is Determined by the DRF Motif in CXCR6 and Is Cell-Type Specific.

CXCR6, the receptor for CXCL16, is expressed on multiple cell types and can be a coreceptor for human immunodeficiency virus 1. Except for CXCR6, all human chemokine receptors contain the D(3.49)R(3.50)Y(3.51) sequence, and all but two contain A(3.53) at the cytoplasmic terminus of the third transmembrane helix (H3C), a region within class A G protein-coupled receptors that contacts G proteins. In CXCR6, H3C contains D(3.49)R(3.50)F(3.51)I(3.52)V(3.53) at positions 126-130. We investigated the importance and interdependence of the canonical D126 and the noncanonical F128 and V130 in CXCR6 by mutating D126 to Y, F128 to Y, and V130 to A singly and in combination. For comparison, we mutated the analogous positions D142, Y144, and A146 to Y, F, and V, respectively, in CCR6, a related receptor containing the canonical sequences. Mutants were analyzed in both human embryonic kidney 293T and Jurkat E6-1 cells. Our data show that for CXCR6 and/or CCR6, mutations in H3C can affect both receptor signaling and chemokine binding; noncanonical H3C sequences are functionally linked, with dual changes mitigating the effects of single mutations; mutations in H3C that compromise receptor activity show selective defects in the use of individual Gi/o proteins; and the effects of mutations in H3C on receptor function and selectivity in Gi/o protein use can be cell-type specific. Our findings indicate that the ability of CXCR6 to make promiscuous use of the available Gi/o proteins is exquisitely dependent on sequences within the H3C and suggest that the native sequence allows for preservation of this function across different cellular environments.

Tuberculosis Chest X-Ray Image Retrieval System Using Deep Learning Based Biomarker Predictions.

The world health organization's global tuberculosis (TB) report for 2022 identifies TB, with an estimated 1.6 million, as a leading cause of death. The number of new cases has risen since 2020, particularly the number of new drug-resistant cases, estimated at 450,000 in 2021. This is concerning, as treatment of patients with drug resistant TB is complex and may not always be successful. The NIAID TB Portals program is an international consortium with a primary focus on patient centric data collection and analysis for drug resistant TB. The data includes images, their associated radiological findings, clinical records, and socioeconomic information. This work describes a TB Portals' Chest X-ray based image retrieval system which enables precision medicine. An input image is used to retrieve similar images and the associated patient specific information, thus facilitating inspection of outcomes and treatment regimens from comparable patients. Image similarity is defined using clinically relevant biomarkers: gender, age, body mass index (BMI), and the percentage of lung affected per sextant. The biomarkers are predicted using variations of the DenseNet169 convolutional neural network. A multi-task approach is used to predict gender, age and BMI incorporating transfer learning from an initial training on the NIH Clinical Center CXR dataset to the TB portals dataset. The resulting gender AUC, age and BMI mean absolute errors were 0.9854, 4.03years and 1.67kgm2. For the percentage of sextant affected by lesions the mean absolute errors ranged between 7% to 12% with higher error values in the middle and upper sextants which exhibit more variability than the lower sextants. The retrieval system is currently available from https://rap.tbportals.niaid.nih.gov/find_similar_cxr.

Evaluation of the synergistic potential and mechanisms of action for de novo designed cationic antimicrobial peptides.

Antimicrobial peptides (AMPs) have emerged as promising candidates in combating antimicrobial resistance - a growing issue in healthcare. However, to develop AMPs into effective therapeutics, a thorough analysis and extensive investigations are essential. In this study, we employed an in silico approach to design cationic AMPs de novo, followed by their experimental testing. The antibacterial potential of de novo designed cationic AMPs, along with their synergistic properties in combination with conventional antibiotics was examined. Furthermore, the effects of bacterial inoculum density and metabolic state on the antibacterial activity of AMPs were evaluated. Finally, the impact of several potent AMPs on E. coli cell envelope and genomic DNA integrity was determined. Collectively, this comprehensive analysis provides insights into the unique characteristics of cationic AMPs.

HASP server: a database and structural visualization platform for comparative models of influenza A hemagglutinin proteins.

BACKGROUND: Influenza A viruses possess RNA genomes that mutate frequently in response to immune pressures. The mutations in the hemagglutinin genes are particularly significant, as the hemagglutinin proteins mediate attachment and fusion to host cells, thereby influencing viral pathogenicity and species specificity. Large-scale influenza A genome sequencing efforts have been ongoing to understand past epidemics and pandemics and anticipate future outbreaks. Sequencing efforts thus far have generated nearly 9,000 distinct hemagglutinin amino acid sequences. DESCRIPTION: Comparative models for all publicly available influenza A hemagglutinin protein sequences (8,769 to date) were generated using the Rosetta modeling suite. The C-alpha root mean square deviations between a randomly chosen test set of models and their crystallographic templates were less than 2 Å, suggesting that the modeling protocols yielded high-quality results. The models were compiled into an online resource, the Hemagglutinin Structure Prediction (HASP) server. The HASP server was designed as a scientific tool for researchers to visualize hemagglutinin protein sequences of interest in a three-dimensional context. With a built-in molecular viewer, hemagglutinin models can be compared side-by-side and navigated by a corresponding sequence alignment. The models and alignments can be downloaded for offline use and further analysis. CONCLUSIONS: The modeling protocols used in the HASP server scale well for large amounts of sequences and will keep pace with expanded sequencing efforts. The conservative approach to modeling and the intuitive search and visualization interfaces allow researchers to quickly analyze hemagglutinin sequences of interest in the context of the most highly related experimental structures, and allow them to directly compare hemagglutinin sequences to each other simultaneously in their two- and three-dimensional contexts. The models and methodology have shown utility in current research efforts and the ongoing aim of the HASP server is to continue to accelerate influenza A research and have a positive impact on global public health.

Analysis of the conformation and function of the Plasmodium falciparum merozoite proteins MTRAP and PTRAMP.

Thrombospondin repeat (TSR)-like domains are structures involved with cell adhesion. Plasmodium falciparum proteins containing TSR domains play crucial roles in parasite development. In particular, the preerythrocytic P. falciparum circumsporozoite protein is involved in hepatocyte invasion. The importance of these domains in two other malaria proteins, the merozoite-specific thrombospondin-related anonymous protein (MTRAP) and the thrombospondin-related apical membrane protein (PTRAMP), were assessed using near-full-length recombinant proteins composed of the extracellular domains produced in Escherichia coli. MTRAP is thought to be released from invasive organelles identified as micronemes during merozoite invasion to mediate motility and host cell invasion through an interaction with aldolase, an actin binding protein involved in the moving junction. PTRAMP function remains unknown. In this study, the conformation of recombinant MTRAP (rMTRAP) appeared to be a highly extended protein (2 nm by 33 nm, width by length, respectively), whereas rPTRAMP had a less extended structure. Using an erythrocyte binding assay, rMTRAP but not rPTRAMP bound human erythrocytes; rMTRAP binding was mediated through the TSR domain. MTRAP- and in general PTRAMP-specific antibodies failed to inhibit P. falciparum development in vitro. Altogether, MTRAP is a highly extended bifunctional protein that binds to an erythrocyte receptor and the merozoite motor.

Predictive capabilities of baseline radiological findings for early and late disease outcomes within sensitive and multi-drug resistant tuberculosis cases.

Purpose: This study compares performance of Timika Score to standardized, detailed radiologist observations of Chest X rays (CXR) for predicting early infectiousness and subsequent treatment outcome in drug sensitive (DS) or multi-drug resistant (MDR) tuberculosis cases. It seeks improvement in prediction of these clinical events through these additional observations. Method: This is a retrospective study analyzing cases from the NIH/NIAID supported TB Portals database, a large, trans-national, multi-site cohort of primarily drug-resistant tuberculosis patients. We analyzed patient records with sputum microscopy readings, radiologist annotated CXR, and treatment outcome including a matching step on important covariates of age, gender, HIV status, case definition, Body Mass Index (BMI), smoking, drug use, and Timika Score across resistance type for comparison. Results: 2142 patients with tuberculosis infection (374 with poor outcome and 1768 with good treatment outcome) were retrospectively reviewed. Bayesian ANOVA demonstrates radiologist observations did not show greater predictive ability for baseline infectiousness (0.77 and 0.74 probability in DS and MDR respectively); however, the observations provided superior prediction of treatment outcome (0.84 and 0.63 probability in DS and MDR respectively). Estimated lung abnormal area and cavity were identified as important predictors underlying the Timika Score's performance. Conclusions: Timika Score simplifies the usage of baseline CXR for prediction of early infectiousness of the case and shows comparable performance to using detailed, standardized radiologist observations. The score's utility diminishes for treatment outcome prediction and is exceeded by the usage of the detailed observations although prediction performance on treatment outcome decreases especially in MDR TB cases.

Analysis, Modeling, and Target-Specific Predictions of Linear Peptides Inhibiting Virus Entry.

Antiviral peptides (AVPs) are bioactive peptides that exhibit the inhibitory activity against viruses through a range of mechanisms. Virus entry inhibitory peptides (VEIPs) make up a specific class of AVPs that can prevent envelope viruses from entering cells. With the growing number of experimentally verified VEIPs, there is an opportunity to use machine learning to predict peptides that inhibit the virus entry. In this paper, we have developed the first target-specific prediction model for the identification of new VEIPs using, along with the peptide sequence characteristics, the attributes of the envelope proteins of the target virus, which overcomes the problem of insufficient data for particular viral strains and improves the predictive ability. The model's performance was evaluated through 10 repeats of 10-fold cross-validation on the training data set, and the results indicate that it can predict VEIPs with 87.33% accuracy and Matthews correlation coefficient (MCC) value of 0.76. The model also performs well on an independent test set with 90.91% accuracy and MCC of 0.81. We have also developed an automatic computational tool that predicts VEIPs, which is freely available at https://dbaasp.org/tools?page=linear-amp-prediction.

Combining Radiological and Genomic TB Portals Data for Drug Resistance Analysis.

Tuberculosis (TB) drug resistance is a worldwide public health problem. It decreases the likelihood of a positive outcome for the individual patient and increases the likelihood of disease spread. Therefore, early detection of TB drug resistance is crucial for improving outcomes and controlling disease transmission. While drug-sensitive tuberculosis cases are declining worldwide because of effective treatment, the threat of drug-resistant tuberculosis is growing, and the success rate of drug-resistant tuberculosis treatment is only around 60%. The TB Portals program provides a publicly accessible repository of TB case data with an emphasis on collecting drug-resistant cases. The dataset includes multi-modal information such as socioeconomic/geographic data, clinical characteristics, pathogen genomics, and radiological features. The program is an international collaboration whose participants are typically under a substantial burden of drug-resistant tuberculosis, with data collected from standard clinical care provided to the patients. Consequentially, the TB Portals dataset is heterogenous in nature, with data representing multiple treatment centers in different countries and containing cross-domain information. This study presents the challenges and methods used to address them when working with this real-world dataset. Our goal was to evaluate whether combining radiological features derived from a chest X-ray of the host and genomic features from the pathogen can potentially improve the identification of the drug susceptibility type, drug-sensitive (DS-TB) or drug-resistant (DR-TB), and the length of the first successful drug regimen. To perform these studies, significantly imbalanced data needed to be processed, which included a much larger number of DR-TB cases than DS-TB, many more cases with radiological findings than genomic ones, and the sparse high dimensional nature of the genomic information. Three evaluation studies were carried out. First, the DR-TB/DS-TB classification model achieved an average accuracy of 92.4% when using genomic features alone or when combining radiological and genomic features. Second, the regression model for the length of the first successful treatment had a relative error of 53.5% using radiological features, 25.6% using genomic features, and 22.0% using both radiological and genomic features. Finally, the relative error of the third regression model predicting the length of the first treatment using the most common drug combination varied depending on the feature type used. When using radiological features alone, the relative error was 17.8%. For genomic features alone, the relative error increased to 19.9%. The model had a relative error of 19.0% when both radiological and genomic features were combined. Although combining radiological and genomic features did not improve upon the use of genomic features when classifying DR-TB/DS-TB, the combination of the two feature types improved the relative error of the predictive model for the length of the first successful treatment. Furthermore, the regression model trained on radiological features achieved the best performance when predicting the treatment length of the most common drug combination.

African Centers of Excellence in Bioinformatics and Data Intensive Science: Building Capacity for Enhancing Data Intensive Infectious Diseases Research in Africa.

Africa faces both a disproportionate burden of infectious diseases coupled with unmet needs in bioinformatics and data science capabilities which impacts the ability of African biomedical researchers to vigorously pursue research and partner with institutions in other countries. The African Centers of Excellence in Bioinformatics and Data Intensive Science are collaborating with African academic institutions, industry partners, the Foundation for the National Institutes of Health (FNIH) and the National Institute of Allergy and Infectious Diseases (NIAID) at the National Institutes of Health (NIH) in a public-private partnership to address these challenges through enhancing computational infrastructure, fostering the development of advanced bioinformatics and data science skills among local researchers and students and providing innovative emerging technologies for infectious diseases research.