Glyco-Enzyme Repository

Home/Data Notes

Data notes and curation worklist

Migrating the site required reading every record mechanically, which exposed places where the original data is internally inconsistent or incomplete. Nothing here has been corrected — the migration preserves the source as it stands. This page lists what a revision pass should look at.

Conflicting identifiers

The legacy site stated each gene's identity twice: in the family index that links to the record, and in the record's own header table. For these 15 values the two disagree in a way that is not merely a naming or formatting difference. One of the two is wrong in each case; the migration does not guess which, and both are shown on the affected record.

Superseded gene symbols

Gene nomenclature has moved on since these records were written. Where a record carries an older symbol alongside the current one — recorded in the source as a parenthesised alternative — both are kept and both are searchable. These are naming differences rather than conflicts, but they mark records whose annotation is likely to be out of date.

Symbols held by more than one record

4 legacy record pages share a gene-symbol value with another page. Each is kept as its own record, named after the source page it came from so the two can be told apart. They need reconciling: either they are two different genes, one of them mislabelled, or they are duplicate entries for one gene.

Records with no CAZy family

114 records carry no usable CAZy family label. Most are the glycan-modifying enzymes CAZy does not classify, which is expected and correct — they are grouped by function instead. A few, listed below with the raw value found in the source, have a malformed entry in that field.

Referenced files that do not exist

36 file references across 13 records point at files the original server did not hold; requesting them returned a 404 there too, so there was nothing to migrate. They are shown on the records as unavailable rather than removed, since the reference is evidence the construct was made.

Records with no orderable clone

1 records describe constructs for which no DNASU clone ID and no Jarvis-lab stock is recorded. Their sequence files are still available. Whether a clone was in fact deposited and simply never recorded here is worth checking against DNASU during the revision.