# QUEX who?…

The University of Queensland and the University of Exeter have partnered to establish the QUEX Institute, a new initiative designed to bolster our joint global research impact.Through the QUEX joint PhD program, you can receive a scholarship to study at two leading universities - The University of Queensland and the University of Exeter - on projects linked with the virtual QUEX Institute of Global Sustainability and Wellbeing. Although a debate is still ongoing on the most correct pronunciation of QUEX (Q-EX or “quecks”, I refuse to use the latter: “quacks”????), I was very fortunate to make part of the first QUEX PhD cohort and so far I’m absolutely loving this experience. You can find out more about QUEX here

One of my direct supervisors is Prof. Naomi Wray, at UQ, that co-heads the Program in Complex Traits Genomics, together with Prof. Peter Visscher and Prof. Jian Yang. I first was introduced to Naomi’s work during one my MSc internships, when I read the paper: “Pitfalls of predicting complex traits from SNPs”. Immediately I thought “these people are great and I’d love to work with them one day”. Fast-forward one and a half years and that brief thought came through. Now my most recurrent thought is “life is great, but I jsut hope I can make through this with at least a fair amount of sanity”. Still love what I do, though.

My other direct supervisor is Prof. Jonathan Mill, at UoE, whose I work I wasn’t as familiar with, simply because I’ve only recently developed an interest in DNA methylation. As I quickly became aware Jon is a household name in the DNAm field despite being super young. Everyone always skeptically asks me for confirmation when I tell them Jon’s my supervisor (“Them: Jon Mill, really? Me: Yes. Them: Oh, great! Btw, he never replies to my emails”). As I’ve been spending more time in Brisbane, I haven’t really had the chance to meet the Exeter team as much as I’d like, but hopefuly I’ll get to that next year!

My other supervisors are Dr. Allan McRae (UQ) and Dr. Eilis Hannon (UoE). They are both great at stats and incredibly smart and I can’t but aspire to be half as great as either of them.

All that being said, this is a GitHub Pages website for hosting my most relevant PhD analyses. Please feel free to comment or send me any suggestions/criticism to m.nabais@uq.edu.au.

# List of most relevant analyses

• Smoking Prediction: Briefly, We aimed to develop a DNA methylation-based predictor of smoking status, using a least absolute shrinkage and selection operator (LASSO) multinomial regression model, to perform variable selection and regularization to enhance predictive accuracy of smoking status, assessed by out-of-sample prediction. Models were fit by 10-fold cross-validation, using CpG methylation values as independent variables and smoking status as response variable; data were taken from the Lothian Birth Cohort 1936 + 1921 (n = 988). $\beta$ > 0 coefficients were extracted and multiplied by raw $\beta$ methylation values of a Motor Neuron Disease cohort (n = 905) giving a methylation index for each individual. Predictive accuracy was then assessed by logistic regression. I presented this as a poster at Genemappers 2018