Single nucleotide polymorphism (SNP)
|
A variation at a single position in a DNA sequence that occurs when one nucleotide is replaced by another in at least 1% of the population. SNPs can result in different alleles at that position. |
Trait
|
A phenotype of an SNP. |
Genome-wide association study (GWAS)
|
A genetic study in which the phenotype is regressed for each genetic variant to correct the bias resulting from multiple testing. |
Genetic variant
|
Any change in the DNA sequence that distinguishes one individual from another. This can include variations such as SNPs. |
Harmonization
|
Some SNPs may have more than one possible allele. Harmonization is the process of formatting GWAS statistics to ensure that exposures and outcomes use the same allele as the effect allele; otherwise, the Mendelian randomization analysis results will be incorrect. |
Palindromic variants
|
An SNP in which the same two nucleotides are found on both forward and reverse strands (e.g., A/T on the forward and T/A on the reverse). This can cause problems in harmonization because it is difficult to establish which is the minor allele in cases where both nucleotides have a similar frequency. |
Mendelian randomization (MR) analysis
|
The statistical analysis used in a two-sample MR design, which estimates the cause association of the exposure on the outcome. |
Multivariable Mendelian randomization (MVMR)
|
A statistical technique that uses genetic variants to investigate the causal relationships between multiple exposure variables and an outcome variable. |
Instruments
|
Instruments were randomly assigned, associated with the exposure, and performed in MR analyses to infer the causal association between exposure and outcome. |
Linkage disequilibrium (LD)
|
A non-random association between nearby genetic variants. In MR studies, LD can lead to biased results if not accounted for. Researchers adjust for LD or select independent variants as instruments to avoid this. |
Proxy variant
|
Some SNPs of exposure cannot be extracted from the dataset of the outcome. We could use other SNPs as proxy variants highly correlated with the unavailable SNPs to perform the MR analysis. |
Genetic liability to the exposure
|
Binary traits with many associated genetic variants have an underlying continuous genetic liability measured through GWASs. People’s genes can make them more or less likely to be affected by environmental factors that can harm them, like pollution or toxins. Some people may have genes that make them more sensitive to these things, while others may have genes that make them less sensitive. |
Pleiotropy
|
Pleiotropy in MR occurs when the genetic instruments used to estimate the causal effect of an exposure on an outcome also directly affect the outcome through other pathways. |
Heterogeneity
|
Heterogeneity indicates a violation of the exclusion restriction assumption in MR (validity), which can lead to biased effect estimates. |
Cochran’s Q statistic
|
A statistical test is used to determine whether there is significant heterogeneity (variation) in the effect sizes of multiple studies included in a meta-analysis. |
Population stratification
|
The uneven distribution of genetic variants between different subpopulations can bias associations between exposures and outcomes in genetic studies. |