Discrepancies in the HIPC Immune Signatures 2 Dataset

HIPC Data Standards
1

Hi,

I’m not sure if this is the right place to post this, but I’ve come across some discrepancies in an ImmuneSpace dataset and thought it would be useful to communicate them here. Specifically, this relates to the HIPC Immune Signatures 2 study on human responses to vaccination and the all_norm_with_response.eset dataset, which contains both transcriptomic and immune response data (accessible here).

In this dataset, antibody measurements are derived from ELISA, neutralizing antibody, or HAI assays. Since individuals may have multiple antibody response measurements at a given timepoint (e.g., for different viral strains or analytes), the dataset is reports the maximum fold change (MFC) for each individual across timepoints compared to pre-vaccination.

The relevant columns include:

  • assay – The assay corresponding to the max fold change measurement
  • maxStrain_MFC – The viral strain or analyte corresponding to the max fold change measurement
  • ImmResp_baseline_timepoint_MFC – The timepoint of the baseline measurement used for the max fold change
  • ImmResp_postVax_timepoint_MFC – The timepoint of the post-vaccination measurement used for the max fold change
  • ImmResp_baseline_value_MFC – The log2 value of the baseline measurement used for the max fold change
  • ImmResp_postVax_value_MFC – The log2 value of the post-vaccination measurement used for the max fold change
  • MFC – The log2 maximum fold change

However, I noticed that for many individuals, the MFC value does not match the difference between ImmResp_postVax_value_MFC and ImmResp_baseline_value_MFC. In fact, out of 1,221 unique individuals with immune response data, I found 359 cases where the reported values were inconsistent.

For example, take participant ID “SUB192189.1325”:

  • The dataset reports neutralising antibodies for the strain Neisseria meningitidis strain A (F8238), with a baseline value of 1, a post-vaccination value of 11, and an MFC of 12.
  • However, cross-checking this individual against the raw neutralising antibody response data (also accessible here) reveals that the reported MFC of 12 actually corresponds to a different strain (Neisseria meningitidis strain C (C11)) with a baseline value of 1 and a post-vaccination value of 13.

Checking additional random individuals showed the same issue: while the MFC values in all_norm_with_response.eset appear correct, the associated columns (assay, maxStrain_MFC, the timepoints etc.) do not always reflect the correct values used to derive the MFC.

Any information on this would be greatly appreciated.
Best,
Arthur