SumStatsRehab: an efficient algorithm for GWAS summary statistics assessment and restoration

Vitamin D and thyroid function: A mendelian randomization study

BACKGROUND

Numerous organs, including the thyroid gland, depend on vitamin D to function normally. Insufficient levels of serum 25-hydroxyvitamin D [25(OH)D] are seen as a potential factor contributing to the emergence of several thyroid disorders, however, the causal relationship remains unclear. Here we use a Mendelian randomization (MR) approach to investigate the causal effect of serum 25(OH)D concentration on the indicators of thyroid function.

METHODS

We conducted a two-sample MR analysis utilizing summary data from the most extensive genome-wide association studies (GWAS) of serum 25(OH)D concentration (n = 443,734 and 417,580), thyroid-stimulating hormone (TSH, n = 271,040), free thyroxine (fT4, n = 119,120), free triiodothyronine (fT3, n = 59,061), total triiodothyronine (TT3, n = 15,829), as well as thyroid peroxidase antibody levels and positivity (TPOAb, n = 12,353 and n = 18,297), low TSH (n = 153,241), high TSH (n = 141,549), autoimmune hypothyroidism (n = 287,247) and autoimmune hyperthyroidism (n = 257,552). The primary analysis was conducted using the multiplicative random-effects inverse variance weighted (IVW) method. The weighted mode, weighted median, MR-Egger, MR-PRESSO, and Causal Analysis Using Summary Effect estimates (CAUSE) were used in the sensitivity analysis.

RESULTS

The IVW, as well as MR Egger and CAUSE analysis, showed a suggestive causal effect of 25(OH)D concentration on high TSH. Each 1 SD increase in serum 25(OH)D concentration was associated with a 12% decrease in the risk of high TSH (p = 0.02). Additionally, in the MR Egger and CAUSE analysis, we found a suggestive causal effect of 25(OH)D concentration on autoimmune hypothyroidism. Specifically, each 1 SD increase in serum 25(OH)D concentration was associated with a 16.34% decrease in the risk of autoimmune hypothyroidism (p = 0.02).

CONCLUSIONS

Our results support a suggestive causal effect which was negative in direction across all methods used, meaning that higher genetically predicted vitamin D concentration possibly lowers the odds of having high TSH or autoimmune hypothyroidism. Other thyroid parameters were not causally influenced by vitamin D serum concentration.

BCFtools/liftover: an accurate and comprehensive tool to convert genetic variants across genome assemblies

MOTIVATION

Many genetics studies report results tied to genomic coordinates of a legacy genome assembly. However, as assemblies are updated and improved, researchers are faced with either realigning raw sequence data using the updated coordinate system or converting legacy datasets to the updated coordinate system to be able to combine results with newer datasets. Currently available tools to perform the conversion of genetic variants have numerous shortcomings, including poor support for indels and multi-allelic variants, that lead to a higher rate of variants being dropped or incorrectly converted. As a result, many researchers continue to work with and publish using legacy genomic coordinates.

RESULTS

Here we present BCFtools/liftover, a tool to convert genomic coordinates across genome assemblies for variants encoded in the variant call format with improved support for indels represented by different reference alleles across genome assemblies and full support for multi-allelic variants. It further supports variant annotation fields updates whenever the reference allele changes across genome assemblies. The tool has the lowest rate of variants being dropped with an order of magnitude less indels dropped or incorrectly converted and is an order of magnitude faster than other tools typically used for the same task. It is particularly suited for converting variant callsets from large cohorts to novel telomere-to-telomere assemblies as well as summary statistics from genome-wide association studies tied to legacy genome assemblies.

AVAILABILITY AND IMPLEMENTATION

The tool is written in C and freely available under the MIT open source license as a BCFtools plugin available at http://github.com/freeseek/score.