Why Non-ASCII Unicode Characters Should Not Be Used in the Human Genome Variation Society Nomenclature

A paper entitled “Solving the :: Fusion Nomenclature Challenge for File and Directory Naming”¹, which was recently published in this journal, proposes alterations to the well-established and globally accepted nomenclature for the formal description of gene fusions² when naming files and directories. A major barrier across life sciences is the widespread presence of incomplete, erroneous, and undiscoverable variant descriptions and associated metadata in publications and databases. Furthermore, additional common characters widely used throughout the HGVS Nomenclature are recommended by the authors to be replaced with Unicode characters that certain typefaces may render as visually similar but are completely distinct.

The authors claim that four characters used in the HGVS variant nomenclature (i.e. :, /, >, and *) are incompatible with file and directory (folder) naming in Windows, Linux and MacOS, which is undoubtedly true. However, they present no cogent argument regarding why files and directories should be named using HGVS-compliant variant descriptions. There are simpler workarounds that preserve any necessary information without violating file and directory naming conventions. The VariantValidator³ interactive variant description validation tool (https://variantvalidator.org/service/validate/) provides a facility to save validation results to a PDF file and creates a filename that is compatible with file-naming conventions for all of the commonly used operating systems. For example, validation of the sequence variant description NM_000088.4:c.589G>T produces a PDF file named “VariantValidator_report_NM_000088.4_c.589G_T.pdf”. In other words, the colon and the greater than symbols are each replaced by underscores and no file naming conventions are violated. This simple character substitution method could similarly be used for replacing / and * characters as well as for directory naming.

To our knowledge, the only variant validation tool capable of identifying and suggesting corrections to Unicode substitutions in variant descriptions is LOVD’s HGVS library (manuscript in preparation; see https://LOVD.nl/HGVS). In accordance with the HGVS Nomenclature guidelines, this library treats Unicode use as invalid and automatically converts such symbols back to the proper ASCII characters. The VariantValidator tools incorporate this LOVD HGVS library. Other, more widely used validation tools fail to process variant descriptions that contain Unicode characters.

The authors make misguided and problematic assertions in their paper such as “…an official update to the HGVS Nomenclature that incorporates these substitutions would help standardize variant representation across both scientific and computational environments.” and that their proposed solution “…promotes harmonization towards a unified genomic variant nomenclature.”

We request the addition of a clear disclaimer warning against the use of any replacement characters in variant descriptions shared or published outside of local systems. If specific characters were replaced with similar looking Unicode characters in file and folder names, these replacements could very easily be propagated to variant descriptions shared in emails, websites, papers, or databases, which would cause major issues.

Current approaches to searching databases and the scientific literature for DNA, RNA, and/or protein variants rely on straightforward text-based matching. Replacing the established ASCII characters with visually similar Unicode characters would break these methods. Moreover, the proposed Unicode symbols are not consistently rendered: they appear correctly only in certain fonts/typefaces. For example, the authors’ article, as first published on sciencedirect.com, illustrated this issue — some of the suggested replacement characters displayed as blank boxes when viewed in Firefox on a Linux system.

The authors’ proposal would disrupt conventions that are established & widely used and could markedly disrupt the uniform reporting of sequence variants with respect to genes, transcripts and proteins. Such changes would damage consistent and reproducible communication of patient data in clinical and non-clinical genetics scenarios. If implemented, the proposed changes could damage established computational systems that seek to identify, compare, extract and correct variant descriptions from various sources, including publications and electronic patient records.

In conclusion, contrary to the authors’ suggestion that the introduction of Unicode characters would “enable compliance with the HGVS Nomenclature”, and “more reliable data exchange, and improved patient care”, the outcome could very likely be the exact opposite. Notably, the authors have now backtracked on their original published stance regarding “harmonization” of the HGVS nomenclature and have amended their README.md file on GitHub (https://github.com/Tomatenbiss/HGVS_CharacterKit). A corresponding warning needs to be published in print for the benefit of those who will never read the online warning.