Remediation of the NMR data of natural glycans

Primary structure elucidation in glycobiology is strongly affected by published structure-reporting NMR signals, especially on the ¹³C nucleus. The glycan NMR simulation accuracy and machine learning outcome depend on the quality of the NMR signal assignment in glycan databases. Within our work on improving the data quality in the Carbohydrate Structure Database (CSDB), we have applied a systematic search for inconsistencies in the published NMR data. The search was based on a bulk comparison between the experimental and simulated ¹³C NMR chemical shifts and manual analysis of the mismatches. On the basis of this analysis, CSDB was remediated by marking and correcting the NMR errors found in 272 structure elucidation reports published over the past 40 years.