Thanks for the great resource! I've found an issue with ieu-b-18, multiple sclerosis.
This dataset was probably manually added by downloading the dataset from https://imsgc.net/?page_id=31. The metadata provided states that it has data on 115000 participants.
When requesting the data, the link to the dataset linked to the 2019 Science paper (https://www.science.org/doi/10.1126/science.aav7188) looks to be only part of the full meta-analysis described in the paper. There is unfortunately no accompanying metadata file, but I am basing this on the following: The paper describes the new data as coming 'from 15 data sets, totaling 14,802 subjects with MS and 26,703 controls for our discovery study.' When comparing the p-values of the dataset in the repository with the p-values in supplementary table S7, it overlaps with the P value for this discovery dataset. The effect estimates for the other datasets used for replication (described by their previous papers, MS chip and MS immunochip) are also provided here, and the overall p-value for the meta-analysis of the discovery dataset + MS chip + immunochip (thus presumably the total 115000 participants) can be found in one of the last columns. Finally, the name of the file (discovery_metav3.0) also points into this direction.
To avoid confusion in choosing the correct GWAS for (MR) analyses, I'd like to request changing the metadata to reflect the actual number of participants as described above.
Thanks for the great resource! I've found an issue with ieu-b-18, multiple sclerosis.
This dataset was probably manually added by downloading the dataset from https://imsgc.net/?page_id=31. The metadata provided states that it has data on 115000 participants.
When requesting the data, the link to the dataset linked to the 2019 Science paper (https://www.science.org/doi/10.1126/science.aav7188) looks to be only part of the full meta-analysis described in the paper. There is unfortunately no accompanying metadata file, but I am basing this on the following: The paper describes the new data as coming 'from 15 data sets, totaling 14,802 subjects with MS and 26,703 controls for our discovery study.' When comparing the p-values of the dataset in the repository with the p-values in supplementary table S7, it overlaps with the P value for this discovery dataset. The effect estimates for the other datasets used for replication (described by their previous papers, MS chip and MS immunochip) are also provided here, and the overall p-value for the meta-analysis of the discovery dataset + MS chip + immunochip (thus presumably the total 115000 participants) can be found in one of the last columns. Finally, the name of the file (discovery_metav3.0) also points into this direction.
To avoid confusion in choosing the correct GWAS for (MR) analyses, I'd like to request changing the metadata to reflect the actual number of participants as described above.