Changelog

The washout timer is the accumulation plotter run the other way: from a last dose to five half-lives, with a row per half-life and the clock time each falls on, seeded from the half-life recorded in a compound's entry. The vial planner turns vial, dose and cadence into the day the vial runs out, and when a compound is named it lays the entry's reconstituted storage window over the same calendar and counts the doses that would fall past it.

Two tools that only work because the corpus records molecular weights. The converter turns micrograms into nanomoles and back using the mass in a compound's entry, so a paper's nmol/kg figure and a label's µg figure can be read against each other. The checker takes the found mass from a certificate's mass-spectrometry line and reports its distance from the theoretical mass in daltons and parts per million, and recognises the one-proton offset of an [M+H]⁺ reading so it is not mistaken for an error.

Naming the compound is not required, but doing so lends two things from its entry: the reconstituted storage line, quoted verbatim, with a discard date counted from the day of mixing when the entry names a number of days; and a printable vial card carrying the concentration, the draw in units and the two dates. Rows from the round-number search now carry a Use button that fills the water in.

A combination vial is its parts sharing the water, so one draw carries every component in the ratio of their masses. The new calculator takes the label's per-component amounts, the water added and a draw in units, and returns what that draw delivers per component — with draws-per-vial alongside. Arithmetic on entered numbers, and the page says where a blend's evidence actually lives: with the components, studied alone.

The calculator gains a second direction: dose known, water unknown. Instead of returning one volume it searches the clean landings — the water amounts that put the dose exactly on 4, 5, 10 units of a U-100 syringe — and offers only exact rows, never rounded ones. 80 mg at 1 mg per dose finds 3.2 mL for a 4-unit draw, and 4 mL for a 5-unit one.

Of 440 ungraded indications, 310 belong to compounds with no citations in the corpus and stay ungraded — there is nothing to link. The other 130 were reviewed by hand against their compound's own citations, catching the synonym matches word-overlap misses: MASH resolution supporting the NASH indication, adipose-mass loss supporting fat loss. 32 cleared the same bar the lexical matcher enforces and now carry computed grades; 98 stayed honestly unlinked. Every authored link is committed with its reason. Graded coverage: 269 to 301 of 709.

This site now counts page views (cookieless, no cross-site identifier) and records what is typed into search along with how many results it returned. Neither carries an IP address, an account or a session identifier, so searches cannot be reassembled into one person's history; search records are deleted after 180 days. The privacy page was rewritten before any of it went live and states the purpose plainly, including the commercial one. Logged here because a reference that publishes its corrections should publish this too.

The command palette indexed only compounds of kind peptide, so the seven named protocols — KLOW, Glow, Wolverine and the rest — plus nine small molecules including MK-677 could not be found by searching, despite having full monographs and inbound links. The index now covers every live monograph and matches on recorded aliases as well as names.

Monographs with a recorded half-life now carry the accumulation model inside the molecular section, pinned to that compound's value — interval switching, steady-state peak and time-to-90% included. Levels are single-dose multiples: a shape, not a pharmacokinetic prediction. The full plotter remains under tools.

Each of the 101 records now carries a generated label illustration in the context rail — one house style, accent color set by category. Every image is captioned as illustrative and is not a product photograph; no vendor, brand or barcode appears on any label.

Every compound's prose — descriptors, sections, indication notes, FAQs, side effects, timelines, checklists, interaction summaries — has been re-expressed in house style. Facts, numbers, doses and findings are unchanged; only the wording is new. Verbatim overlap with the upstream source fell from 99.6% to 1%, and what remains is bare factual labels (disease names, contraindication labels) with a single honest phrasing. 24 FAQ entries that restated content available elsewhere on their page were removed.

A conservative matcher now joins each indication to its compound's own citations, matching only on distinctive shared terms weighted by rarity, above a strict threshold. Every applied link is committed with its earning terms to a reviewable audit file. 269 of 709 indications carry linked, graded references (95 at human-clinical grade); the rest stay ungraded rather than borrowing the compound's strongest study.

Every PMID-bearing citation is now typed by study design from PubMed's own metadata (155 of 223 settled by MeSH publication types and descriptors; the rest are labelled as inferred). Each compound carries the published P/H/R axes: preclinical depth and human evidence are computed from the typed references, regulatory standing is authored with its authority cited, and a validator recomputes the axes on every grading run and warns on drift. Indications stay ungraded until their own references are linked.

Optional accounts shipped: bookmark compounds to a personal library, save stack layouts, delete everything from settings. Owner-scoped row-level security; the privacy page was rewritten to present tense in the same change.

thepeptidereference.com went live with every desktop screen resolving: browse, categories, best-for rankings, monographs, three calculators, references, guides, methodology and legal.

101 compound records ingested end to end: 223 citations, 709 research indications, 565 interaction edges, 63 recorded half-lives. Sparse entries render as sparse rather than padded.

The upstream source records lysine at position 2; PubChem CID 9941957 records glutamate (GEPPPGKPADDAGLV). The monograph follows PubChem, and the correction is pinned in the ingest layer so re-imports cannot regress it.

Two PMIDs resolved on PubMed to unrelated papers and were deleted outright. An entry with fewer, checkable references beats one padded with citations that do not say what they claim.

The BPC-157 monograph became the template spine: ten sections, sequence map, evidence panel, and a verification gate that checks the rendered page against the design contract.