[Recorded Webinar and Slides are Available Now!] Join dkNET Webinar: Solving the Undiagnosed Diseases through Machine Learning

*Watch recorded webinar here: https://youtu.be/ypN44EHRFRI

*Webinar slides: 
https://www.slideshare.net/dkNET/dknet-webinar-solving-the-undiagnosed-diseases-through-machine-learning-05142021

Join dkNET Webinar on Friday, May 14, 2021, 11 am - 12 pm PDT

Abstract

Every year hundreds of patients face uncertainty when healthcare providers are unable to discover the cause for their symptoms. The Undiagnosed Diseases Network (UDN) is a research study backed by the National Institutes of Health Common Fund that seeks to provide answers for patients and families affected by these mysterious conditions. For patients with potential rare genetic disorders, sequencing will be performed to identify the disease-causing variant. The process of defining pathogenicity currently requires labor-intensive manual searches of a variety of databases and web resources. This manual process is time-consuming, subject to inter-user variability and variations in the depth or quality of the databases. It also requires broad expertise across multiple biological and informatics domains. Here, we created a systematic, comprehensive search engine, MARRVEL (Model organism Aggregated Resources for Rare Variant ExpLoration, http://marrvel.org), that mines all the critical information for variant analysis and presents it in a succinct, user-friendly way. MARRVEL integrates human databases (OMIM, gnomAD, ExAC, ClinVar, Geno2MP, DGV, and DECIPHER) and seven model organism databases from yeast to mammals. Furthermore, we are also developing a Knowledge-based and Explainable Artificial Intelligent system (MARRVEL-AI) to prioritize and identify novel disease-causing coding variants. The interpretability of a machine learning method inversely correlates with its accuracy for complex tasks. To circumvent this, we are combining different models of artificial intelligence with complementary strengths, such as expert system and random forest. With only a small training data set, our model achieved a high accuracy in identifying disease causing variants for UDN cases.

Top. 3 key questions that Undiagnosed Disease Network (UDN) can answer:

1. Which gene/its variants is likely to be the cause of a rare Mendelian Disorder?

2. What is the probable disease mechanism?

3. Do we have an animal model for the rare disease?

Presenter: Zhandong Liu, PhD, Associate Professor, Department of Pediatrics, Baylor College of Medicine

Dial-in Information: 

Date/Time: Friday, May 14, 2021, 11 am - 12 pm PDT
https://uchealth.zoom.us/meeting/register/tZMrcOmhqDoiGNIkJ5VoIxvoUUmP10PULKst

Upcoming webinars schedule: https://dknet.org/about/webinar