BUG in Nightingale pipeline
Closed this issue · 1 comments
The current code depends on information contained within a file 'ng_anno' within the package itself to identify 'derived' features in the Nightingale Health NMR data. However, where Nightingale datasets have additional metabolites/features not in the list provided and/or metabolites/features are named differently, this part of the pipeline fails. As a consequence, steps reliant on this information including the generation of metabolite/feature and sample missingness stats and PCA, and subsequent filtering based on these are not working as expected.
We recommend extreme caution when using this pipeline whilst we investigate this issue. It may be possible to avoid the problem by specifying 'Other' in place of 'Nightingale' in the parameter file but this approach has not been fully tested at this time.
Issue resolved with new Nightingale annotation file that sits within the package (ng_anno), along with integration of some extra checks to flag when metabolite and/or sample IDs in the different elements (sample metadata, feature metadata, metabolite data, ng_anno) do not match. See 'Known Issues' in README.