Autor: |
Liu, Xueping, Nudel, Ron, Thompson, Wesley K., Appadurai, Vivek, Schork, Andrew J., Buil, Alfonso, Rasmussen, Simon, Allesøe, Rosa L., Werge, Thomas, Mors, Ole, Børglum, Anders D., Hougaard, David M., Mortensen, Preben B., Nordentoft, Merete, Benros, Michael E. |
Jazyk: |
angličtina |
Rok vydání: |
2021 |
Zdroj: |
Liu, X, Nudel, R, Thompson, W K, Appadurai, V, Schork, A J, Buil, A, Rasmussen, S, Allesøe, R L, Werge, T, Mors, O, Børglum, A D, Hougaard, D M, Mortensen, P B, Nordentoft, M & Benros, M E 2021, ' Corrigendum to “Genetic factors underlying the bidirectional relationship between autoimmune and mental disorders – Findings from a Danish population-based study” [Brain Behav. Immun. 91 (2021) 10–23] (Brain Behavior and Immunity (2021) 91 (10–23), (S0889159120300799), (10.1016/j.bbi.2020.06.014)) ', Brain, Behavior, and Immunity, vol. 96, pp. 307-308 . https://doi.org/10.1016/j.bbi.2021.05.019 |
Popis: |
We have discovered an error pertinent to this paper, which occurred during the preparation of the marker dataset used in the four GWAS analyses described in our paper. We emphasize, however, that the overall result and conclusions of the GWAS analyses remain the same, and all the other analyses in the paper are unaffected by this error. In this corrigendum we specify which parts of the Results are affected and to what extent. The original text stated that the dataset previously used by Schork et al. (Schork et al., 2019) was used as the starting point for the SNP-based analyses in the paper. However, a different dataset, which should have gone through similar quality control (QC), was used. We have only recently discovered an error that occurred during the removal of some of the markers that had failed the preliminary QC measures. We summarize the preliminary and study-specific QC steps at the end of this corrigendum. The starting point for the analyses in this study was the raw dosage dataset, from which, due to an error, only some of the markers which had failed the preliminary QC were removed. As stated in the paper, however, during the GWAS procedure markers with INFO score < 0.8 and MAF < 0.01 in the respective GWAS samples were excluded from downstream analysis. Thus, the majority of markers which had failed the preliminary QC steps were removed by the above two filters, but not all of them. This applied only to the dataset used in the GWASs. The polygenic risk score (PRS) analyses used the QCed hard calls dataset, with the preliminary QC steps (as described below), and with individual genotypes with probability −34), which shows an effect similar to that of the previous top SNP. Additionally, this SNP is correlated with the previous top SNP (r2 = 0.45, p < 0.0001, χ2 = 89.2835, using LDlink with the CEU population). The top SNP on chromosome 7 in the secondary association peak with the comorbid diagnosis remains the same. Regarding the Index SNPs (Table 4), the top SNP in our original paper (rs1064173) was flagged only for missingness (>10%), and one other SNP (rs3957147) was flagged for missingness and for the HWE test. We note that the missingness filter was based on missingness post imputation, introduced during the removal of specific imputed genotypes which had a probability score 10% of subjects, where the missingness in imputed genotypes was determined by treating individual genotypes with probability < 0.9 as missing for the purpose of this check; Hardy-Weinberg equilibrium (HWE) P < 1 × 10−6 (in controls); a significant association with the genotyping wave. Although not previously mentioned, a small number of markers showed association with the imputation batch itself, and these were also removed, as were markers with differential missingness across psychiatric cases and controls (P < 1 × 10−6). In a more stringent QC, employed when only best-guess genotypes, or hard calls, were used, only genotypes with probability of 0.9 or above were retained, and markers were retained if they had an INFO score (following the imputation) of at least 0.8 and MAF of at least 0.01 (in the QCed sample). The study-specific QC in this paper employed the INFO score and MAF thresholds as per the stringent QC, but these were from the output of each respective GWAS. |
Databáze: |
OpenAIRE |
Externí odkaz: |
|