Scott Kulm Graduate Student

About Me Projects

My Expertise

Current Graduate Student at Weill Cornell Medicine, coadvised by Drs. Elemento and Mezey. Researching polygenic risk scores, and in particular how pleiotropy can be leveraged to improve disease prediction. I like to live at the cross roads of coding, statistics, biology and creative thought. Former chemical engineering student at The Cooper Union, stil a son and brother.

Learn

As a graduate student I am always actively attempting to pick up a new skill, whether it be based in statistics, coding or biological.

Build

So others may implement the lessons I've learned, I attempt to code up some tools or simple exemplars on how coding can work

Teach

Personally sharing knowledge has always been deeply rewarding, the lessons I upload here can hopefully do the same on a much larger scale

Featured Projects

prs_methods

Creation of Polygenic Risk Score Reports

Doctors working in clinical settings could advantageously utilize polygenic risk scores. However, the details on how polygenic risk scores are communicated and what they imply are still largely open question. With a new genotyping initiative at my institution I explore those questions here.

Check it out


assumptions

Disease Risk Model Assumptions

Polygenic risk scores are judged in a model framework. The exact model employed, and specifically the assumptions of that model, have a large impact on the accuracy estimation of the corresponding score.

Check it out


prs_methods

Performance of GWAS Statistic Adjustment Methods

Polygenic risk scores are generated from external genome-wide summary statistics that are adjusted for the purpose of disease risk estimation. Multiple methods are available to make the adjustment, and it is not clear which method works best.

Check it out


rare_variant

Cardiac Pathogenic Rare Variants

Rare variants identified with whole exome sequencing have a large yet underappreciated effect on many diseases, including cardiac conditions. In this project I walk through the identification of pathogenic variants and show how they increase the risk of atrial fibrillation.

Check it out


qgg

Quantitative Genetics Notes

For the past semester I have been a TA in the Cornell course Quantiative Genetics and Genomics. For the class I ran computer lab, making tutorial-like R-Markdowns that teach both coding and the fundamentals of conducting a GWAS

Check it out


ldpred

LDPred Derivation

I have been working to fill in the missing links within the method LDPred, which shrinks effect sizes in polygenic risk score calculations. I have found a few errors, so please give a read and see if I am right.

Check it out


ticTacToe

Tic Tac Toe Three Ways

I think tic tac toe is an ideal way to learn about a programming language - so I have coded up tic tac toe programs of varying complexity in three different languages: R, Python, and Julia. If you know only one language you can use these examples as a Rosetta Stone to language learning.

Check it out