Week 2: Slight Change of Plans

Caitlin E -

Hello Everyone! This week, I gained a clearer understanding of what my project is about, as well as what my responsibilities will be for the next few weeks. Although my research still involves data analysis in the genetic diagnoses of neurodevelopmental conditions, the exact data analysis models I will be using differ from what I originally anticipated. I will be analyzing the results of two different AI models– AI MARRVEL (AIM) and ClinPrior– to uncover the genetic bases of neurological conditions. Implementing these tools can improve the accuracy and efficiency of genetic diagnoses. 

What is AIM? 

AI MARRVEL, or simply MARRVEL (Model organism Aggregated Resources for Rare Variant ExpLoration),  is an artificial intelligence system designed to help diagnose rare genetic disorders. This model employs a random-forest machine-learning classifier, a model commonly used for classification. The model derives information from several databases including ClinVar and gnomAD. With training on over 3.5 million genetic variants from thousands of diagnosed cases, AIM’s unique features make it an effective tool in discovering molecular diagnoses. 

What is ClinPrior? 

ClinPrior is a data analysis tool that uses an interactome-based prioritization approach to rank candidate causal genetic variants based on a patient’s phenotype, meaning that ClinPrior matches phenotypic information with possible genetic diagnoses through a multi-layered evaluation method. The prioritization algorithm can be broken into three steps: the calculation of a phenotypic association metric by comparing patient phenotypes to data in existing human disease databases, the propagation of the phenotypic score within a gene-gene metric, and the variant filtration of Variant Call Format file and subsequent calculation of variant deleteriousness score. This three-step process links a certain trait, condition, or symptom to a genetic sequencing difference before finally predicting whether the genetic variant caused a specific phenotype. Through this processing sequence, ClinPrior helps maximize diagnostic yield from Whole-exome Sequencing and Whole-genome Sequencing, which is especially important for rare disease cases. 

Next Steps 

For the rest of this week (and possibly next week), I will be working remotely to interpret results from the two AI tools. The data sets that the AI examined mainly focus on neurodevelopmental diseases that result in severe seizures. The genetic diagnoses for some of these diseases have already been found, but most of the cases remain unsolved for now. I hope to share some findings soon! 

-Caitlin

More Posts

Comments:

All viewpoints are welcome but profane, threatening, disrespectful, or harassing comments will not be tolerated and are subject to moderation up to, and including, full deletion.

    kira_a
    Hi Caitlin! I find your project fascinating and I enjoy the thorough explanations for each concept you cover in your blogs. I think the work you are doing in the field of neurodevelopmental conditions is very important and I was wondering what you plan to do with your work, or what kind of applications it will have when you are done with the project. Do you plan on creating a final product or are you more focused on the process of the research itself? I can't wait to read more as you progress further in your work!
    caitlin_e
    Hi Kira! Thank you for your interest! Although I am more focused on the process of the research itself, I am working on finalizing at least a few of the genetic diagnoses in the dataset I am evaluating. In the timespan we have for the senior project, I do not foresee officially publishing my findings before the final presentation, but I remain intent on doing what I can to analyze results.

Leave a Reply

Your email address will not be published. Required fields are marked *