Week 03

Topics

This week’s assignments will guide you through the following topic:

eQTL (expression quantitative trait loci) mapping to understand genetic regulatory effects on gene expression.

Please read the following:

Landmark paper for cis-eQTL mapping across human tissues (Read the whole paper - it is very integral to our replication project)
Identification of cis-eQTLs in single cell data (new model considerations) (Just take a look at the methods section and compare to the models used in the GTEx paper)
Regarding formatting a scientific paper (Just take a look at how the paper is structured, e.g. abstract, introduction, methods, results, discussion)

Using the 1000Genomes genotype data and this gene expression data, match the sample IDs of the individuals between the two datasets.
Run an eQTL analysis for a single gene of your chosing. For this gene, select only the SNPs within +/- 500 kb of the gene body (you can use gene_annot.txt.gz as the master list of all genes genome-wide). You can select these SNPs by using plink with the –extract bed1 flag. For example: plink2 –bfile 1000G.EUR.15 –extract bed1 $my_coord –out 1000G.EUR.15.GeneX –make-bed (in this case, $my_coord is a single line 6-column tab delimited text file with the genomic coordinates of the +/- 500 kb window around your gene of interest, for example, chr15 200000 1200000 mygene 0 +
You can use different software to run an eQTL analysis. For example, fastQTL is a user-friendly software although I haven’t used it before. I prefer to use simple linear regression models in R or python and use a for loop or an apply() function. In R, you can use the lm() function as follows: lm(y_gene_expression ~ genotype_snp_j). If you use R, you can use the plink2R library and the read_plink function to read a .bed plink file of genotypes: genos = read_plink(“1000G.EUR.15.bed”,impute=”avg”)
Bonus: to visualize the results of an eQTL analysis, we like to plot the -log10 p-value of the linear regression (y-axis) versus the genomic coordinate of the SNP (x-axis). This is known as a locusZoom plot (http://locuszoom.org/).

Answer the following questions on Gradescope: