Dataset Information

A novel machine learning-based approach for the computational functional assessment of pharmacogenomic variants.

ABSTRACT:

Background

The field of pharmacogenomics focuses on the way a person's genome affects his or her response to a certain dose of a specified medication. The main aim is to utilize this information to guide and personalize the treatment in a way that maximizes the clinical benefits and minimizes the risks for the patients, thus fulfilling the promises of personalized medicine. Technological advances in genome sequencing, combined with the development of improved computational methods for the efficient analysis of the huge amount of generated data, have allowed the fast and inexpensive sequencing of a patient's genome, hence rendering its incorporation into clinical routine practice a realistic possibility.

Methods

This study exploited thoroughly characterized in functional level SNVs within genes involved in drug metabolism and transport, to train a classifier that would categorize novel variants according to their expected effect on protein functionality. This categorization is based on the available in silico prediction and/or conservation scores, which are selected with the use of recursive feature elimination process. Toward this end, information regarding 190 pharmacovariants was leveraged, alongside with 4 machine learning algorithms, namely AdaBoost, XGBoost, multinomial logistic regression, and random forest, of which the performance was assessed through 5-fold cross validation.

Results

All models achieved similar performance toward making informed conclusions, with RF model achieving the highest accuracy (85%, 95% CI: 0.79, 0.90), as well as improved overall performance (precision 85%, sensitivity 84%, specificity 94%) and being used for subsequent analyses. When applied on real world WGS data, the selected RF model identified 2 missense variants, expected to lead to decreased function proteins and 1 to increased. As expected, a greater number of variants were highlighted when the approach was used on NGS data derived from targeted resequencing of coding regions. Specifically, 71 variants (out of 156 with sufficient annotation information) were classified as to "Decreased function," 41 variants as "No" function proteins, and 1 variant in "Increased function."

Conclusion

Overall, the proposed RF-based classification model holds promise to lead to an extremely useful variant prioritization and act as a scoring tool with interesting clinical applications in the fields of pharmacogenomics and personalized medicine.

SUBMITTER: Pandi MT

PROVIDER: S-EPMC8351412 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundPersonalized asthma management depends on a clinician's ability to efficiently review patient's data and make timely clinical decisions. Unfortunately, efficient and effective review of these data is impeded by the varied format, location, and workflow of data acquisition, storage, and processing in the electronic health record. While machine learning (ML) and clinical decision support tools are well-positioned as potential solutions, the translation of such frameworks requires that barriers to implementation be addressed in the formative research stages.ObjectiveWe aimed to use a structured user-centered design approach (double-diamond design framework) to (1) qualitatively explore clinicians' experience with the current asthma management system, (2) identify user requirements to improve algorithm explainability and Asthma Guidance and Prediction System prototype, and (3) identify potential barriers to ML-based clinical decision support system use.MethodsAt the "discovery" phase, we first shadowed to understand the practice context. Then, semistructured interviews were conducted digitally with 14 clinicians who encountered pediatric asthma patients at 2 outpatient facilities. Participants were asked about their current difficulties in gathering information for patients with pediatric asthma, their expectations of ideal workflows and tools, and suggestions on user-centered interfaces and features. At the "define" phase, a synthesis analysis was conducted to converge key results from interviewees' insights into themes, eventually forming critical "how might we" research questions to guide model development and implementation.ResultsWe identified user requirements and potential barriers associated with three overarching themes: (1) usability and workflow aspects of the ML system, (2) user expectations and algorithm explainability, and (3) barriers to implementation in context. Even though the responsibilities and workflows vary among different roles, the core asthma-related information and functions they requested were highly cohesive, which allows for a shared information view of the tool. Clinicians hope to perceive the usability of the model with the ability to note patients' high risks and take proactive actions to manage asthma efficiently and effectively. For optimal ML algorithm explainability, requirements included documentation to support the validity of algorithm development and output logic, and a request for increased transparency to build trust and validate how the algorithm arrived at the decision. Acceptability, adoption, and sustainability of the asthma management tool are implementation outcomes that are reliant on the proper design and training as suggested by participants.ConclusionsAs part of our comprehensive informatics-based process centered on clinical usability, we approach the problem using a theoretical framework grounded in user experience research leveraging semistructured interviews. Our focus on meeting the needs of the practice with ML technology is emphasized by a user-centered approach to clinician engagement through upstream technology design.

Project description:Introduction: Video-based clinical rating plays an important role in assessing dystonia and monitoring the effect of treatment in dyskinetic cerebral palsy (CP). However, evaluation by clinicians is time-consuming, and the quality of rating is dependent on experience. The aim of the current study is to provide a proof-of-concept for a machine learning approach to automatically assess scoring of dystonia using 2D stick figures extracted from videos. Model performance was compared to human performance. Methods: A total of 187 video sequences of 34 individuals with dyskinetic CP (8-23 years, all non-ambulatory) were filmed at rest during lying and supported sitting. Videos were scored by three raters according to the Dyskinesia Impairment Scale (DIS) for arm and leg dystonia (normalized scores ranging from 0-1). Coordinates in pixels of the left and right wrist, elbow, shoulder, hip, knee and ankle were extracted using DeepLabCut, an open source toolbox that builds on a pose estimation algorithm. Within a subset, tracking accuracy was assessed for a pretrained human model and for models trained with an increasing number of manually labeled frames. The mean absolute error (MAE) between DeepLabCut's prediction of the position of body points and manual labels was calculated. Subsequently, movement and position features were calculated from extracted body point coordinates. These features were fed into a Random Forest Regressor to train a model to predict the clinical scores. The model performance trained with data from one rater evaluated by MAEs (model-rater) was compared to inter-rater accuracy. Results: A tracking accuracy of 4.5 pixels (approximately 1.5 cm) could be achieved by adding 15-20 manually labeled frames per video. The MAEs for the trained models ranged from 0.21 ± 0.15 for arm dystonia to 0.14 ± 0.10 for leg dystonia (normalized DIS scores). The inter-rater MAEs were 0.21 ± 0.22 and 0.16 ± 0.20, respectively. Conclusion: This proof-of-concept study shows the potential of using stick figures extracted from common videos in a machine learning approach to automatically assess dystonia. Sufficient tracking accuracy can be reached by manually adding labels within 15-20 frames per video. With a relatively small data set, it is possible to train a model that can automatically assess dystonia with a performance comparable to human scoring.

Project description:Background: During the last decades a number of genome-wide association studies (GWASs) has identified numerous single nucleotide polymorphisms (SNPs) associated with different complex diseases. However, associations reported in one population are often conflicting and did not replicate when studied in other populations. One of the reasons could be that most GWAS employ a case-control design in one or a limited number of populations, but little attention was paid to the global distribution of disease-associated alleles across different populations. Moreover, the majority of GWAS have been performed on selected European, African, and Chinese populations and the considerable number of populations remains understudied. Aim: We have investigated the global distribution of so far discovered disease-associated SNPs across worldwide populations of different ancestry and geographical regions with a special focus on the understudied population of Armenians. Data and Methods: We have used genotyping data from the Human Genome Diversity Project and of Armenian population and combined them with disease-associated SNP data taken from public repositories leading to a final dataset of 44,234 markers. Their frequency distribution across 1039 individuals from 53 populations was analyzed using self-organizing maps (SOM) machine learning. Our SOM portrayal approach reduces data dimensionality, clusters SNPs with similar frequency profiles and provides two-dimensional data images which enable visual evaluation of disease-associated SNPs landscapes among human populations. Results: We find that populations from Africa, Oceania, and America show specific patterns of minor allele frequencies of disease-associated SNPs, while populations from Europe, Middle East, Central South Asia, and Armenia mostly share similar patterns. Importantly, different sets of SNPs associated with common polygenic diseases, such as cancer, diabetes, neurodegeneration in populations from different geographic regions. Armenians are characterized by a set of SNPs that are distinct from other populations from the neighboring geographical regions. Conclusion: Genetic associations of diseases considerably vary across populations which necessitates health-related genotyping efforts especially for so far understudied populations. SOM portrayal represents novel promising methods in population genetic research with special strength in visualization-based comparison of SNP data.