Exploring the Relationship between Spatial Transcriptomics and Metal Accumulation in Amyotrophic Lateral Sclerosis

Panav Mhatre, Vismay Ravikumar & Hrishikesh Deosthali

Lay Summary:

By exploring the relatonship between genes and metals that may be potentially neurotoxic, we observed multiple biological mechanisms that may be involved in the progression of ALS. We are seeking to train a machine learning pipeline we developed on the dataset at our disposal to predict metal abundance from tissue imaging and gene expression data, aiding doctors with diagnoses of ALS.

Abstract:

Currently, elemental imaging requires LA-ICP TOF-MS laser ablation imaging, a resource that is neither widely available nor affordable. Accumulations of metals such as manganese, zinc[1], and copper[2] have been associated with ALS, and thus elemental imaging is a potential avenue for diagnoses and understanding the genetic and biochemical pathways underlying the disease. This paper explores the relationship between tissue morphology, spatial transcriptomic data, and metal accumulation in amyotrophic lateral sclerosis(ALS) cases. To ensure proper alignment between all modalities of data, Hematoxylin and eosin(H&E) stains of brain stem tissue in ALS patients were coregistered with corresponding metal accumulation maps and spatial transcriptomics data containing 18,031 genes collected using Visium. Then, a nearest neighbors algorithm was run on the metal maps to determine the Visium spot to which each pixel corresponded. After averaging the metal accumulation values assigned to each Visium spot, Spearman correlation analysis was run on the top 47 genes, to which a p_value was obtained. Enrichment pathway analysis was run for each metal to determine which genetic pathways the isolated genes were related to.

Q&A:

Bios: Panav Mhatre,Vismay Ravikumar,Hrishikesh Deosthali

Program Track: Advanced Research

GitHub Username:

PanavMhatre -Panav Mhatre

vizdiz -Vismay Ravikumar

rishydeosthali -Hrishikesh Deosthali

What was your favorite seminar? Why?

The seminar with Jason Wei because it was super cool to see his work at Google with prompting and also his work at OpenAI. -Panav Mhatre

My favorite seminar was Gokul’s seminar on ST inference as his presentation was thorough and engaging, and I had been interested in his research since early on in the internship when I read his preprint article on the Cell2RNA framework. -Vismay Ravikumar

My favorite seminar was on the the multimodal analysis of metals, Spatial Transcriptomics, and Histological Structures in colorectal cancer given by Aruesha. Since our group worked with the same type of data, I think this seminar appealed to me especially through the lens of learning about the metal signaling and transport pathway analysis done. Definitely was a really interesting seminar. -Hrishikesh Deosthali

If you were to summarize your summer internship experience in one sentence, what would it be?

I loved the ability to be able to work on real-world applications of ML in the pathalogy space rather than like a tutorial-based program. -Panav Mhatre

Learning from my mentors, teammates, and literature, while applying what I learned to make new discoveries. -Vismay Ravikumar