Sequence Similarity Network Analysis Provides Insight into the Temporal and Geographical Distribution of Mutations in SARS-CoV-2 Spike Protein.

Severe acute respiratory syndrome-related coronavirus (SARS-CoV-2), which still infects hundreds of thousands of people globally each day despite various countermeasures, has been mutating rapidly. Mutations in the spike (S) protein seem to play a vital role in viral stability, transmission, and adaptability. Therefore, to control the spread of the virus, it is important to gain insight into the evolution and transmission of the S protein. This study deals with the temporal and geographical distribution of mutant S proteins from sequences gathered across the US over a period of 19 months in 2020 and 2021. The S protein sequences are studied using two approaches: (i) multiple sequence alignment is used to identify prominent mutations and highly mutable regions and (ii) sequence similarity networks are subsequently employed to gain further insight and study mutation profiles of concerning variants across the defined time periods and states. Additionally, we tracked the variants using visualizations on geographical maps. The visualizations produced using the Directed Weighted All Nearest Neighbors (DiWANN) networks and maps provided insights into the transmission of the virus that reflect well the statistics reported for the time periods studied. We found that the networks created using DiWANN are superior to commonly used approximate distance networks created using BLAST bitscores. The study offers a richer computational approach to analyze the transmission profile of the prominent S protein mutations in SARS-CoV-2 and can be extended to other proteins and viruses.

Collaborative Multi-Expert Active Learning for Mobile Health Monitoring: Architecture, Algorithms, and Evaluation.

Mobile health monitoring plays a central role in the future of cyber physical systems (CPS) for healthcare applications. Such monitoring systems need to process user data accurately. Unlike in other human-centered CPS, in healthcare CPS, the user functions in multiple roles all at the same time: as an operator, an actuator, the physical environment and, most importantly, the target that needs to be monitored in the process. Therefore, mobile health CPS devices face highly dynamic settings generally, and accuracy of the machine learning models the devices employ may drop dramatically every time a change in setting happens. Novel learning architecture that specifically address challenges associated with dynamic environments are therefore needed. Using active learning and transfer learning as organizing principles, we propose a collaborative multiple-expert architecture and accompanying algorithms for the design of machine learning models that autonomously adapt to a new configuration, context, or user need. Specifically, our architecture and its constituent algorithms are designed to manage heterogeneous knowledge sources or experts with varying levels of confidence and type while minimizing adaptation cost. Additionally, our framework incorporates a mechanism for collaboration among experts to enrich their knowledge, which in turn decreases both cost and uncertainty of data labeling in future steps. We evaluate the efficacy of the architecture using two publicly available human activity datasets. We attain activity recognition accuracy of over 85 % (for the first dataset) and 92 % (for the second dataset) by labeling only 15 % of unlabeled data.

A nearest-neighbors network model for sequence data reveals new insight into genotype distribution of a pathogen.

BACKGROUND: Sequence similarity networks are useful for classifying and characterizing biologically important proteins. Threshold-based approaches to similarity network construction using exact distance measures are prohibitively slow to compute and rely on the difficult task of selecting an appropriate threshold, while similarity networks based on approximate distance calculations compromise useful structural information. RESULTS: We present an alternative network representation for a set of sequence data that overcomes these drawbacks. In our model, called the Directed Weighted All Nearest Neighbors (DiWANN) network, each sequence is represented by a node and is connected via a directed edge to only the closest sequence, or sequences in the case of ties, in the dataset. Our contributions span several aspects. Specifically, we: (i) Apply an all nearest neighbors network model to protein sequence data from three different applications and examine the structural properties of the networks; (ii) Compare the model against threshold-based networks to validate their semantic equivalence, and demonstrate the relative advantages the model offers; (iii) Demonstrate the model's resilience to missing sequences; and (iv) Develop an efficient algorithm for constructing a DiWANN network from a set of sequences. We find that the DiWANN network representation attains similar semantic properties to threshold-based graphs, while avoiding weaknesses of both high and low threshold graphs. Additionally, we find that approximate distance networks, using BLAST bitscores in place of exact edit distances, can cause significant loss of structural information. We show that the proposed DiWANN network construction algorithm provides a fourfold speedup over a standard threshold based approach to network construction. We also identify a relationship between the centrality of a sequence in a similarity network of an Anaplasma marginale short sequence repeat dataset and how broadly that sequence is dispersed geographically. CONCLUSION: We demonstrate that using approximate distance measures to rapidly construct similarity networks may lead to significant deficiencies in the structure of that network in terms centrality and clustering analyses. We present a new network representation that maintains the structural semantics of threshold-based networks while increasing connectedness, and an algorithm for constructing the network using exact distance measures in a fraction of the time it would take to build a threshold-based equivalent.

Synthetic Sensor Data Generation for Health Applications: A Supervised Deep Learning Approach.

Recent advancements in mobile devices, data analysis, and wearable sensors render the capability of in-place health monitoring. Supervised machine learning algorithms, the core intelligence of these systems, learn from labeled training data. However, labeling vast amount of data is time-consuming and expensive. Moreover, sensor data often contains personal information that a user may not be comfortable sharing. Therefore, there is a strong need to develop methods for generating realistic labeled sensor data. In this paper, we propose a supervised generative adversarial network architecture that learns from feedback from both a discriminator and a classifier in order to create synthetic sensor data. We demonstrate the effectiveness of the architecture on a publicly available human activity dataset. We show that our generator learns to output diverse samples that are similar but not identical to the training data.

Personalized Human Activity Recognition using Wearables: A Manifold Learning-based Knowledge Transfer.

Human activity recognition (HAR) is an important component in health-care systems. For example, it can enable context-aware applications such as elderly care and patient monitoring. Relying on a set of training data, supervised machine learning algorithms form the core intelligence of most existing HAR systems. Meanwhile, the accuracy of an HAR model highly depends on the similarity between the training and the operating context. Therefore, there is a need for developing machine learning algorithms that can easily adapt to the operating context at hand. In this paper, we propose a cross-subject transfer learning algorithm that links source and target subjects by constructing manifolds from feature-level representation of the source subject(s). Our algorithm assigns labels to the unlabeled data in the current context using the manifold learned from the source subject(s). The newly labeled data is used to develop a personalized HAR model for the current context (i.e., target subject). We demonstrate the efficacy of the algorithm using a publicly available dataset on HAR. We show that the proposed framework improves the accuracy of activity recognition by up to 24%.

Characterization of Anaplasma marginale subsp. centrale Strains by Use of msp1aS Genotyping Reveals a Wildlife Reservoir.

Bovine anaplasmosis caused by the intraerythrocytic rickettsial pathogen Anaplasma marginale is endemic in South Africa. Anaplasma marginale subspecies centrale also infects cattle; however, it causes a milder form of anaplasmosis and is used as a live vaccine against A. marginale There has been less interest in the epidemiology of A. marginale subsp. centrale, and, as a result, there are few reports detecting natural infections of this organism. When detected in cattle, it is often assumed that it is due to vaccination, and in most cases, it is reported as coinfection with A. marginale without characterization of the strain. A total of 380 blood samples from wild ruminant species and cattle collected from biobanks, national parks, and other regions of South Africa were used in duplex real-time PCR assays to simultaneously detect A. marginale and A. marginale subsp. centrale. PCR results indicated high occurrence of A. marginale subsp. centrale infections, ranging from 25 to 100% in national parks. Samples positive for A. marginale subsp. centrale were further characterized using the msp1aS gene, a homolog of msp1α of A. marginale, which contains repeats at the 5' ends that are useful for genotyping strains. A total of 47 Msp1aS repeats were identified, which corresponded to 32 A. marginale subsp. centrale genotypes detected in cattle, buffalo, and wildebeest. RepeatAnalyzer was used to examine strain diversity. Our results demonstrate a diversity of A. marginale subsp. centrale strains from cattle and wildlife hosts from South Africa and indicate the utility of msp1aS as a genotypic marker for A. marginale subsp. centrale strain diversity.

RepeatAnalyzer: a tool for analysing and managing short-sequence repeat data.

BACKGROUND: Short-sequence repeats (SSRs) occur in both prokaryotic and eukaryotic DNA, inter- and intragenically, and may be exact or inexact copies. When heterogeneous SSRs are present in a given locus, we can take advantage of the pattern of different repeats to genotype strains based on the SSRs. Cataloguing and tracking these repeats can be difficult as diverse groups of researchers are involved in the identification of the repeats. Additionally, the task is error-prone when done manually. RESULTS: We developed RepeatAnalyzer, a new software tool capable of tracking, managing, analysing and cataloguing SSRs and genotypes using Anaplasma marginale as a model species. RepeatAnalyzer's analysis capability includes novel metrics for measuring regional genetic diversity (corresponding to variety and regularity of SSR occurrence). As a part of its visualization capabilities, RepeatAnalyzer produces high quality maps of the geographic distribution of genotypes or SSRs over a region of interest. RepeatAnalyzer's repeat identification functionality was validated for all SSRs and genotypes reported in 21 publications, using 380 A. marginale isolates gathered from the five publications within that list that provided access to their isolates. The tool produced accurate genotyping results in every case. In addition, it uncovered a number of errors in the published literature: 11 cases where SSRs were misreported, 5 cases where two different SSRs had been given the same name, and 16 cases where two or more names had been given to a single SSR. The analysis and visualization functionalities of the tool are demonstrated using several examples. CONCLUSIONS: RepeatAnalyzer is a robust software tool that can be used for storing, managing, and analysing short-sequence repeats for the purpose of strain identification. The tool can be used for any set of SSRs regardless of species. When applied to A. marginale, our test case, we show that genotype lengths for a given region follow a normal distribution, while SSR frequencies follow a power-law-like distribution. Further, we find that over 90 % of repeats are 28 to 29 amino acids long, which is in agreement with conventional wisdom. Lastly, our analysis reveals that the most common edit distance is five or six, which is counter-intuitive since we expected that result to be closer to one, resulting from the simplest change from one repeat to another.