gps2gtfs
gps2gtfs spent a release making its docs stop describing functions it does not have.
A side-by-side editorial comparison of ducksemantics and glyenzy — release velocity, themes, recent moves, and the top alternatives to consider.
ducksemantics puts an ontology graph and ColBERT retrieval inside DuckDB, callable from R.
ducksemantics is a first release, and an unusually complete one. It combines a DuckDB-native semantic graph, lexical grounding, dense retrieval, exact late-interaction scoring, and structured judgment behind a single R API. The release notes report a full HPO ontology load of 19,836 nodes with 202,740 materialized is_a closure rows, plus persisted EmbeddingGemma and LFM2.5-ColBERT vectors, with DuckDB MaxSim agreeing with an external implementation to seven decimal places.
Glycan biosynthesis as a traceable enzyme graph, now including sulfation and gaps it can bridge.
glyenzy infers which enzymes could have produced a glycan and traces biosynthetic routes to it, backed by curated per-enzyme rules for human glycosyltransferases and, since 0.7.0, twelve sulfotransferases. Biosynthesis functions return typed network objects that keep their igraph interface while supporting layered DAG plots with glycan nodes and labelled enzyme edges. Where no concrete enzyme covers a step, bounded virtual transitions bridge the gap and are marked so users can see which edges are inferred rather than enzymatic.
ducksemantics is a first release, and an unusually complete one. It combines a DuckDB-native semantic graph, lexical grounding, dense retrieval, exact late-interaction scoring, and structured judgment behind a single R API. The release notes report a full HPO ontology load of 19,836 nodes with 202,740 materialized is_a closure rows, plus persisted EmbeddingGemma and LFM2.5-ColBERT vectors, with DuckDB MaxSim agreeing with an external implementation to seven decimal places.
With one release on the record there is no trend to read, only a stated posture: correctness first. The release documents transactional and idempotent graph writes, atomic and resumable embedding caches, index preservation across graph projection, and a verified numerical match against a reference MaxSim implementation. That is the profile of a package expecting to be used as infrastructure rather than a demo.
Where this goes next is not yet determinable from a single release. The obvious pressure point is ontology coverage beyond the HPO used for validation, but the entries do not indicate a plan.
glyenzy infers which enzymes could have produced a glycan and traces biosynthetic routes to it, backed by curated per-enzyme rules for human glycosyltransferases and, since 0.7.0, twelve sulfotransferases. Biosynthesis functions return typed network objects that keep their igraph interface while supporting layered DAG plots with glycan nodes and labelled enzyme edges. Where no concrete enzyme covers a step, bounded virtual transitions bridge the gap and are marked so users can see which edges are inferred rather than enzymatic.
Two kinds of release alternate here. One is enzyme curation, a steady stream of rule corrections for the FUT, MAN1A and MGAT families and removals where an enzyme turned out to act only on glycolipids, which is the unglamorous accuracy work a rule-based inference engine lives on. The other is turning biosynthesis output into a first-class object: paths became networks, networks became typed with plotting support, and targets became a marked vertex attribute. The package moves in lockstep with its siblings, pinning glyrepr 0.13.0 and glymotif 0.17.0 as those refreshed their data and matching APIs, and the latest release already speaks glydraw 0.8.0's orientation values.
The paucimannose N-glycan support dropped in 0.7.0 is the obvious loose end, with users told to stay on 0.6.3, so a reinstated implementation is a plausible next move. Beyond that the virtual-step machinery is new enough that its heuristics, particularly the inferred step limits added in 0.8.1, should keep being tuned.
Other Analytics products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either ducksemantics or glyenzy.
gps2gtfs spent a release making its docs stop describing functions it does not have.
dvir keeps making disaster victim identification a single call instead of a workflow.
pedbuildr reconstructs pedigrees from DNA, and it just got much faster at the search.
forrel is getting faster at the simulations forensic kinship work actually spends its time on.
pedFamilias exists to read one legacy file format, and it has that job nearly finished.
pedmut turns awkward mutation models into ones the likelihood engine can actually handle.
See all ducksemantics alternatives → · See all glyenzy alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. glyenzy is currently shipping more aggressively (velocity 6.3 vs 2.5), with 1 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. glyenzy is currently shipping more aggressively (velocity 6.3 vs 2.5), with 1 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other Analytics products to evaluate alongside.
Top ducksemantics alternatives in Analytics are ranked by recent ship velocity. Browse the "ducksemantics alternatives" section above for the current picks, or visit /alternatives/ducksemantics for the full list with editorial commentary on each.
Top glyenzy alternatives in Analytics are ranked by recent ship velocity. Browse the "glyenzy alternatives" section above for the current picks, or visit /alternatives/glyenzy for the full list with editorial commentary on each.