Small-molecule linking
Joins two seed compounds with a fragment-database-sourced linker, using CReM over a real ChEMBL fragment database. Use for fragment-based drug discovery — combining two smaller candidates into one, not extending or modifying a single lead.
Small-molecule linking
What it does
Takes two seed compounds and proposes candidates that join them with a linker fragment — the fragment-based-drug-discovery question of combining two smaller pieces into one candidate, distinct from Small-molecule generation (extends one seed) and Small-molecule optimization (modifies part of one seed).
Each candidate comes back with QED, SA score, and similarity to each of the two seeds — a well-formed linked product should still resemble each parent in a different part of the molecule, so both numbers are worth checking, not just one.
How candidates are generated
Same underlying fragment-database matching the other two tools use, applied to finding a real, chemically-observed linker that can bridge both seeds at once. Works best on small, fragment-sized seeds — joining two already-large candidate compounds tends to produce an oversized result that scores poorly on drug-likeness; this tool is built for the fragment-linking scale its algorithm actually answers, not for combining two finished leads.
If no linker can be found at the most specific chemical-context match, the tool automatically tries a more permissive match before giving up — and says so in a warning when that happens, along with which setting it used.
Reading the output
Similarity to A and similarity to B together tell the story a single similarity number can't — a candidate that's 0.6 similar to A and 0.1 similar to B is barely using B at all, which may or may not be what's wanted. QED and SA score work the same way they do for the other two tools.
Input
Two seed compounds, both required, each saved as an SDF.
Important limitations
No pocket-awareness, same as the other two tools.
No guaranteed minimum number of results. Some seed pairs have little compatible chemistry to bridge them, even after the automatic retry — a small or empty result set can be a real, correct answer about the chemistry, not a sign something went wrong.
QED and SA score are heuristics, not predictions — see Small-molecule generation's own doc for the same caveat, which applies here identically.