LoginSign Up
← Back to all sparks
G

glyparse

ANALYTICS
Velocity0.0

Parsing Glycan Structure Text Representations

github.com/glycoverse/glyparse ↗

glyparse is quietly becoming the universal reader for glycan text notation.

glycomicsparsingfile formatsperformancer packages
◆Current state
glyparse converts the various text encodings of glycan structures into glyrepr objects. Version 0.7.0 was the widest release in the window, adding parsers for GlyCAM IUPAC, IUPAC-compact, KCF, and LINUCS and teaching auto_parse() to route them, alongside a long list of GlycoCT and WURCS edge cases. The rest of the window is error-handling and performance work: an on_failure switch, a 10-30x speedup on large inputs, and NA and name preservation.
◆Where it's heading
Two things are being built at once. Coverage keeps widening, both in the number of notations supported and in how much malformed or ambiguous real-world input each parser tolerates: alditol residues, unknown ring positions, ambiguous sialic acid descriptors, uppercase residue IDs. Meanwhile the package is hardening for batch use, with progress bars, vectorized speed, and a choice between erroring and returning NA on unparsable strings.
◆Prediction
With the major notations now covered, expect further releases to go to robustness on messy inputs and to keeping pace with glyrepr's structure representation rather than adding formats.

◆Recent moves

  1. 1mo ago

    Parser output uses glyrepr's public constructor

    Switching to glyrepr's public structure constructor fixes tidyverse joins failing on parsed structures under glyrepr 0.13.0. A one-line compatibility fix, but it unblocked a common downstream pattern.

    View source ↗
  2. 1mo ago

    Four new notations parsed and auto-detected

    GlyCAM IUPAC, IUPAC-compact, KCF, and LINUCS each get a parser, and auto_parse() detects and routes them, so callers no longer need to identify their encoding. The same release absorbs a batch of GlycoCT and WURCS edge cases and adds opt-in progress bars for large inputs.

    View source ↗
  3. 1mo ago

    WURCS and GlycoCT edge cases handled

    Ambiguous residues, unknown ring closures, N-sulfate substituent codes, and unknown reducing-end ring positions are handled correctly. Narrow fixes that only surface on real database exports.

    View source ↗
  4. 3mo ago

    on_failure lets parsers return NA instead of erroring

    An on_failure parameter on every parser chooses between erroring and returning NA for unparsable strings, defaulting to the old behavior. Small in surface, but it is what makes parsing a large vector of mixed-quality strings practical.

    View source ↗
  5. 5mo ago

    IUPAC parser accepts plain-text equivalents

    parse_iupac_extended() accepts alpha, beta, and -> spellings in place of their symbols. Input tolerance for structures that passed through systems that mangled Unicode.

    View source ↗
  6. 5mo ago

    Parsers get 10-30x faster on large inputs

    A 10-30x speedup across the parser functions for large inputs, plus removal of a deprecated dplyr call. Parsing is the first step of every glycoverse pipeline, so this is felt on every run.

    View source ↗