MUII Logo University of Missouri

Events

Showing 1-10 of 73

Event Image

2206A Student Center

Monday, Apr. 10th, 2017
11:00AM-12:00PM

share article

Computationally-Determined FliC Linked Recognition Epitope Micelles for Burkholderia Vaccination

Published on 04/06/2017

Human disease caused by Burkholderia spp. is a serious problem in many parts of the world including infection of immunocompromised patients and those with cystic fibrosis.  Traditional antimicrobial therapy is protracted and problematic. The goal of this cooperated structural vaccinology project is to identify novel immune dominant and cryptic linked B and T cell epitopes for the development of efficacious vaccines against Burkholderia mallei and B. pseudomallei flagellar protein FliC. For our in silico part works, we will screen the greater than 30 known genomic/proteomic data sets of Burkholderia spp. including both pathogenic and non-pathogenic strains.  Identified sequences of interest will be scored for inclusion of both a putative T cell (linear) epitope and a B cell (linear or conformational) epitope while eliminating potentially self-reactive mimetic epitopes.  Then candidate FliC peptides will be synthesized and tested by cooperated labs on campus.

Event Image

S110 Memorial Union

Monday, Apr. 3rd, 2017
11:00AM-12:00PM

share article

Dynamically Predict Risk of 30-day Hospital Readmission for Diabetic Patients Using a Data Stream Mining Approach

Published on 03/29/2017

Early hospital readmissions can negatively impact patients’ quality of life and hospitals’ income. Intensive efforts have been made to develop 30-day hospital readmission risk prediction models. Most of the reported statistical and machine learning models were built with static datasets. The reality, however, is that data arrives sequentially and the trend may evolve over time. As a result, many static models failed in practice. To make the models up-to-date, the traditional approach is re-training them from scratch periodically, which can be expensive. Most importantly, it is difficult to determine and discard outdated knowledge from the models using this approach. In this work, we aim to develop a self-adaptive 30-day hospital readmission prediction model using a data stream mining approach. We used a diabetic inpatient encounter data stream to dynamically train and evaluate models based on incremental learning algorithms. This approach and preliminary results will be presented in the seminar.

Event Image

2206A Student Center

Monday, Mar. 20th, 2017
11:00AM-12:00PM

share article

Informatics Approaches to Uncover the Molecular Mechanisms of Endometriosis in Clinical Patients

Published on 03/15/2017

Endometriosis, a complex and common gynecological disorder affecting 5–10% of reproductive-age women, is characterized by the growth of endometrial tissue outside of the uterine cavity. Accumulating evidence indicates that various epigenetic aberrations are associated with endometriosis. In our study, we have methylation data and clinical information on 80 patients (36 controls and 44 cases) from a clinical study. Our objectives are to identify the genomic regions associated with endometriosis and identify the specific genes associated with endometriosis after adjusting for potential confounding variables. We have developed a bioinformatics methylation data analysis pipeline in-house using several open source tools including FastQC, Bowtie2 and R packages. In this seminar, the informatics approach for the data analysis will be presented.

Event Image

2206A Student Center

Monday, Mar. 13th, 2017
11:00AM-12:00PM

share article

A Deep Neural Network Method for Predicting Mitochondrially Localized Proteins in Plants

Published on 03/13/2017

Targeting and translocation of proteins to the appropriate subcellular compartments is crucial for cell organization and function. Some newly synthesized proteins are transported to mitochondria with the assistance of complex targeting sequences containing either an N-terminal pre-sequence or a multitude of internal signals. Compared with experimental approaches, computational predictions provide an efficient way to infer subcellular localization for any given protein. However, it is still challenging to predict plant mitochondrially localized proteins accurately due to various limitations. Consequently, the performance of current tools is unsatisfactory. We present a novel computational approach for large-scale prediction of plant mitochondrial proteins. We collected subcellular localization data for plant proteins from databases and the literature, and extracted different types of features from the training data, including amino acid composition, protein sequence profile, and gene co-expression information. We then trained deep neural networks for predicting plant mitochondrial proteins. Benchmarked on an independent dataset, our method achieves considerable improvements over existing tools in predicting mitochondria-localized proteins in plants. We improved the true positive rate by 10-30% over three of the state-of-the-art tools under similar specificity levels. We also applied our method to predict candidate mitochondrial proteins on the whole proteome of Arabidopsis and potato.

Event Image

2206A Student Center

Monday, Mar. 6th, 2017
11:00AM-12:00PM

share article

Genomic Selection using Deep Learning method

Published on 03/05/2017

Genomic selection is an approach to enhance the quantitative traits in plant and animal breeding program at early stage using whole genome molecular markers, especially for long life-cycle species. It’s based on the assumption that all quantitative trait loci (QTL) tend to be in linkage disequilibrium with at least on marker. Statistical methods, such as ridge regression, best linear unbiased prediction (RR-BLUP)[1], Bayes A[2], Bayesian LASSO[3] are widely used for genomic selection problem works SNP matrix.  Other machine learning methods (random forrest, support vector machine and neural network)[4] are also been applied for this study. In this work, we are developing a deep learning method using long short term memory (LSTM) recurrent network on a public standard dataset of Pinus taeda (loblolly pine) [5]. The stem height trait (HT, cm) was measured across 861 individuals genotyped with 4,853 SNPs derived from 32 parents. The genomic estimated breeding values (GEBV) was calculated using 10-fold cross-validation method and accuracy was measured using Pearson correlation coefficient between GEBV and observed values.

 

Reference:

[1] Hoerl, Arthur E., and Robert W. Kennard. "Ridge regression: biased estimation for nonorthogonal problems." Technometrics 42.1 (2000): 80-86.

[2] Meuwissen, T.H.E.,  B.J. Hayes, and M.E. Goddard. 2001. Prediction of total genetic value using genome-wide dense marker maps. Genetics 157:1819–1829.

[3] Park, Trevor, and George Casella. "The bayesian lasso." Journal of the American Statistical Association 103.482 (2008): 681-686.

[4] Heslot, Nicolas, et al. "Genomic selection in plant breeding: a comparison of models." Crop Science 52.1 (2012): 146-160.

[5] Resende, Márcio FR, et al. "Accuracy of genomic selection methods in a standard data set of loblolly pine (Pinus taeda L.)." Genetics 190.4 (2012): 1503-1510.

Event Image

2205A Student Center

Monday, Feb. 20th, 2017
11:00AM-12:00PM

share article

A pilot study of ePHR implementation impact on physician workflow

Published on 02/16/2017

The ePHR (electronic Personal Health Record) is a self-service technology (SST) used in health care, which can serve as an electronic information source for patients, physician and the government. Based on the literature review, we found that there are some concerns and barriers during the ePHR implement in industry perspectives, physicians’ perspective, patients’ perspectives and technology perspectives. In the pilot study of ePHR implementation impact on physician workflow, we conduct a qualitative analysis using structured physician interviews, and a quantitative analysis for physician workflow observations. We try to create recommendations for ePHR implementation for a variety of physician practices, at the same time, discussed some other important ePHR implementation components: communication, training and measurement.

Event Image

2205A Student Center

Monday, Feb. 13th, 2017
01:00PM-02:00PM

share article

MUII Dissertation Defense - Lincoln Sheets

Published on 02/12/2017

Informatics Strategies for Risk Stratification in Population Health Management

 

Risk analysis and population health management can improve health outcomes, but improved risk stratification is needed to manage healthcare costs. Analysis of 157 publications on translational implementations of “risk stratification in population health management of chronic disease” showed a consensus that population health management and risk stratification can improve outcomes, but found uncertainty over best methods for risk prediction and controversy over the cost savings. The consensus of another 85 publications on the methodologies of “data mining for predictive healthcare analytics” was that clinically interpretable machine learning techniques are more appropriate than “black box” techniques for structured big data sources in healthcare, and the “area under the curve” of a prediction model’s sensitivity versus one-minus-specificity is a standard and reliable way to measure the model’s discrimination. This study used clinically interpretable machine-learning algorithms, combined with simple but powerful data analytic techniques such as cost analysis and data visualization, to evaluate and improve risk stratification for a managed patient population.

This study retrospectively observed 10,000 mid-Missouri Medicare and Medicaid patients between 2012 and 2014. Cost and utilization analyses, statistical clustering, contrast mining, and logistic regression were used to identify patients within a managed population at risk for higher healthcare costs, demonstrate longitudinal changes in risk stratification, and characterize detailed differences between high-risk and low-risk patients. The two highest risk stratification tiers comprised only 21% of patients but accounted for 43% of prospective charges. Patients in the most expensive sub-cluster of the most expensive risk tier were nearly twice as costly as high-risk patients on average. Combining contrast mining with logistic regression predicted the most expensive 5% of patients with 84% accuracy, as measured by area under the curve.

All the strategies used in this study, from the simplest to the most sophisticated, produced useful results. By predicting the small number of patients who will incur the majority of healthcare expenses in terms that are clinically interpretable, these methods can support population health managers in focusing preventive and longitudinal care more effectively. These models, and similar models developed by integrating diverse informatics strategies, could improve health outcomes, delivery, and costs. 

Event Image

2206A Student Center

Monday, Feb. 13th, 2017
11:00AM-12:00PM

share article

Decoding the auxin code: uncovering new players in the auxin pathway through WGCNA analysis

Published on 02/10/2017

Auxins are a class of phytohormones in plants which have an active role in growth and development. The Auxin hormone control pathway in maize meristem is not well studied and has significant scope for in-depth exploration. Previous studies have not shown much novel information about how Auxin is regulated. Weighted Gene co-expression analysis takes results from differentially expressed genes and organizes them into clusters and modules showing possible interactions and co-regulation. These clusters usually highlight unique interactions which cannot be seen by most other methods. We present our ongoing work in building gene co-regulatory networks using certain gene knockouts in the maize meristem Auxin network.

Event Image

2206A Student Center

Monday, Feb. 6th, 2017
11:00AM-12:00PM

share article

RDF-Based Method to Uncover Implicit Health Communication Episodes from Unstructured Health Data

Published on 02/03/2017

Health communication is the process that coordinates health services such as specimen transaction, oral interactions, medical records, and more. Healthcare workflows are based on communication established historically through the practice of healthcare or by the leadership in health institutions. However, during healthcare practices communication doesn’t flow according to plan; interpersonal miscommunication, technical glitches, information overload, etc. risk inefficient healthcare services. We hypothesize that health records contain information related to communication and we can retrieve it in order to address issues of communication. We present an informatics pipeline to retrieve health communication episodes from unstructured health data. The method uses Resource Description Framework (RDF), ontological modeling, and description logic inference to uncover and quantify implicit communication episodes. Retrieved communication has the potential to optimize and improve health communication structures especially, in data-intensive to precision medicine settings.

Event Image

2206A Student Center

Monday, Jan. 30th, 2017
11:00AM-12:00PM

share article

Comparison of histone patterns in orthologus regions between mouse and human tissues

Published on 01/24/2017

With the advent of next-generation sequencing technologies, a considerable effort has been put into sequencing the epigenome of different species. The efforts such as “Encode” and “Roadmap” epigenomics projects provide an opportunity to compare epigenomes across species (especially between human and mouse). This study is an effort to understand how different histone modifications vary/co-appear between orthologus regions of the two species. In this work, we have also used various measures of orthologus similarity between each pair of orthologus genes and explore how histone modifications are conserved with respect to changes in these similarity measures. These measures of similarity include “gene ontology semantic similarity” (GOSemsim), “codon usage frequency similarity” (CUFS), Ka/Ks ratio and gene expression similarity. Our simulations indicate that evolutionary selection pressure of an orthologus pair (Ka/Ks ratio) is more strongly correlated with its histone modification than any other similarity measure.

Showing 1-10 of 73

Back to top