Search the library

Plain words work. Use the arrow keys to move, Enter to open, Esc to close.

Menu

Reference·2026·Living style guide

ADTF design system 1.2. Every part, shown working.

The foundation has to be three things at once: an institute that makes a claim and proves it, a journal that holds twenty years of papers, and a guide for a family that has just heard a diagnosis. The system gives each a register and a surface, and holds them together with five signatures. Every token and component on this page is drawn by the same stylesheet the site uses.

Version
1.0
Date
10 November 2026
Replaces
v0.3
Source of truth
system.css
SectionContents

Values need scripts

Every specimen on this page renders without scripts. The values printed beside them (hex codes, sizes, contrast ratios) are read from the live stylesheet by a script, so they cannot drift from system.css. Turn scripts on to see them.

Principles01

Three registers, held together by five signatures.

In brief

  1. Say only what is true. Every number carries its source or its date. An example is tagged as one.
  2. Plain words first. Lead with the question a parent would ask. The scientific term comes second.
  3. Phone first, calm always. Designed at 360px, usable from the keyboard and with scripts off. Nothing moves unless the reader moves it.

A page alternates registers: a sparse statement, then a dense ledger, then a task. White is the dominant surface. The five signatures are what make a page recognisably ADTF. Each is shown working here and specified in the sections that follow.

  1. Signature 1

    Two voices

    Fira Sans Condensed makes claims, asks questions and names things. Fira Sans is the interface and the data. Serif italic in the voice colour is the second voice: at most one phrase in a section.

    .voice.journal

    Paper·2018·Clinical trial

    We find what is treatable in autism. Then we prove it.

    Frye RE, Slattery J, Delhey L, et al. Molecular Psychiatry, 2018

  2. Signature 2

    One mark, one thing

    Data is drawn as unit charts. One dot is one real paper, one child in a trial, one project. Countable, honest, and never a decorative chart.

    .units.u.u--off.legend

    Assigned leucovorin, 23Assigned placebo, 25

    The 2018 trial. One mark is one child.Source: ADTF statement, 1 October 2025.
  3. Signature 3

    Square by default

    Buttons, fields, cards, panels, photos and figures have 2 to 6px corners. Round is kept for the three shapes where round is the affordance: the search field, filter chips and switches.

    --r-content--r-control--r-tool--r-pill

    Square: everything you read or press

    See the shape rulesPlain summaryOpen access

    Round: only search, chips and switches

  4. Signature 4

    Orange is for giving

    The give button and the fill of a funding meter, and nothing else. Not links, not highlights, not “buy”. Inside the giving widget only the final button is orange.

    .btn--give.meter

    $104,000 raised of $260,000Example

  5. Signature 5

    Every figure is captioned and sourced

    A chart gets “Fig. N”, a caption and a source. A number gets a date or a source. An example value is tagged so nobody mistakes it for a real one.

    .fig__cap.stat__src.placeholder-tag

    293 children in 5 blinded, controlled trials of leucovorin ADTF statement, 1 October 2025
Registers and surfaces02

One composition, three registers.

A surface class remaps the semantic tokens. Below, one composition is rendered four times: only the class on its container changes.

Paper·2018·Clinical trial

The first blinded trial of leucovorin in autism. Then four more.

5 blinded, controlled trials and 293 children. Read the 2018 trial.

Frye RE, Slattery J, Delhey L, et al. Molecular Psychiatry, 2018 Plain summary

Paper .s-paper (the default). The journal register: the library, papers, long reading, evidence. Dense and typographic. White is the dominant surface of every page.

Paper·2018·Clinical trial

The first blinded trial of leucovorin in autism. Then four more.

5 blinded, controlled trials and 293 children. Read the 2018 trial.

Frye RE, Slattery J, Delhey L, et al. Molecular Psychiatry, 2018 Plain summary

Mist .s-mist. The guide register: paths for families, tools, tasks, courses. Plain. The only tint in the system: there are no green, yellow or peach sections.

Paper·2018·Clinical trial

The first blinded trial of leucovorin in autism. Then four more.

5 blinded, controlled trials and 293 children. Read the 2018 trial.

Frye RE, Slattery J, Delhey L, et al. Molecular Psychiatry, 2018 Plain summary

Deep .s-deep. The institute register: statements, proof, the opening of a page, giving moments. Sparse: one large serif claim, luminous data, a portrait.

Paper·2018·Clinical trial

The first blinded trial of leucovorin in autism. Then four more.

5 blinded, controlled trials and 293 children. Read the 2018 trial.

Frye RE, Slattery J, Delhey L, et al. Molecular Psychiatry, 2018 Plain summary

Abyss .s-abyss. The deepest step, for the footer and the utility bar only. It is never a section of a page.

At most two deep bands on a page

DoOpen on deep, give one chapter to deep, and let white carry the rest. Alternate density: a statement, then a ledger, then a task.
Do notA page that is half dark reads as a start-up, and a page of coloured bands reads as a template. Never three sections of the same density in a row.
Colour03

Teal-ink carries the brand. Orange is kept for giving.

There are two layers. The raw palette names the colours. The semantic tokens say what a colour is for, and each surface remaps them. Components read only the semantic layer, which is why they work on paper, mist and deep without change.

The raw palette

Never used directly in a component

  • --inkText and brand on light surfaces
  • --ink-2Secondary text
  • --ink-3Meta text and labels
  • --paperThe default surface
  • --mistThe guide surface; sunk panels on paper
  • --mist-2Sunk panels on mist; empty marks
  • --lineHairlines on paper
  • --deepThe statement surface
  • --abyssThe footer: the deepest step
  • --raiseA raised panel on deep
  • --tealLinks and data marks on light
  • --teal-brightData marks on deep
  • --aquaSecond voice and links on deep
  • --on-deepText on deep
  • --on-deep-2Secondary text on deep
  • --on-deep-3Labels on deep
  • --sunHighlighter; focus ring on deep
  • --sun-50Caution and placeholder wash
  • --giveGiving on light surfaces
  • --give-deepGiving on hover; text on the giving wash
  • --give-brightGiving on deep surfaces
  • --give-50Giving wash, for a tag only
  • --okStatus: done, funded
  • --warnStatus: take care
  • --errorStatus: a field error

Status colours never carry meaning alone: each sits beside a word or an icon.

Semantic tokens, by surface

.s-paper.s-mist.s-deep.s-abyss

TokenJob
--bgThe surface
--bg-raiseA raised panel: tool card, panel
--bg-sunkA sunk panel: in brief, note, citation, tag
--fgText
--fg-2Secondary text
--fg-3Meta text and labels
--ruleHairlines
--rule-strongThe section bar; the top of a tool card
--linkLinks and text actions
--voiceThe serif-italic second voice
--markA data mark
--mark-2The ring of a comparison mark
--mark-offA dimmed mark; the track of a meter
--mark-hlThe one highlighted mark
--btn-bgPrimary button; a chosen chip or option
--btn-fgIts label
--btn-bg-hoverPrimary button on hover
--give-bgThe give button; the fill of a meter
--give-fgIts label
--give-bg-hoverThe give button on hover
--field-bgFields and chips
--field-borderTheir border
--focusThe focus ring
--hlThe highlighter

Each cell names the raw colour the token resolves to on that surface, read from the page as you see it.

Contrast, computed live

WCAG 2.2: text 4.5:1 (AA), 7:1 (AAA); large text and non-text 3:1

The ratios below are computed in the browser from the colours in use, so they need scripts.

PairPaperMistDeepAbyss
Fixed colours, the same on every surfaceResult

The orange rule

.btn--give.meter

$104,000 raised of $260,000Example

DoOrange on the button that gives and on the fill of a real funding meter. Everything else on the page is ink and teal.
Buy on AmazonRead the 2018 trial

New·2025

A tinted band around a call to give.

Do notNo orange links, labels or highlights, no orange “buy” button (buying is not giving), and no tinted band behind the widget.
Typography04

A serif for what we claim, a sans for what you use.

Two families, both self-hosted variable fonts. Every size is a token; the larger ones grow with the viewport. The readouts beside each specimen are measured from this page at its current width.

The two families

--serif--sans

Fira Sans Condensed

Claims, questions, titles and long reading. Light (300) for display, regular (400) for titles and reading, italic for the second voice and for journal names. Never bold.

Fira Sans

Interface, labels, data and body text. Regular (400) for text, medium (500) for navigation, semibold (600) for buttons, labels and names.

The scale

size / line-height · weight, measured now

.display-hero--t-hero Home hero only

Then we prove it.

h1 · .display-1--t-d1 The one page title

Moving treatments from laboratories to lives.

h2 · .display-2--t-d2 Section statements; quotes

Five blinded trials, 293 children.

h3 · .h3--t-h3 Titles of cards and groups

Could my child’s brain be low in folate?

h4 · .h4--t-h4 Sans headings inside components

What the trials measured

h5 · .h5The smallest heading

Reviewed by a clinician

.lead--t-lead One paragraph under a heading

Twenty years of laboratory and clinical research into folate, mitochondria and the immune system, tested in blinded trials and explained for families.

.lead--serifA lead in the first voice

Leucovorin is a prescription form of folate. It is not an “autism pill”.

.prose--t-read Long-form reading

Dr. Frye has studied leucovorin in autism for about two decades.

body--t-body Never below 17px

Leucovorin needs the right patient, dose and formulation. Decisions about it belong with your child’s doctor.

.small--t-small Supporting lines

Sold by Metabolic Learning Resource, founded by Dr. Frye.

.caption · .meta--t-cap Captions and sources. The floor.

Frye RE, Slattery J, Delhey L, et al. Molecular Psychiatry, 2018

.kicker--t-label The label: the only capitals

Paper·2018·Clinical trial

.folioSection numbers

02

The two voices

.voicecite.journal

A question, in serif

Why does my child lose skills when they are sick?

A tagline, in the second voice

Moving treatments from laboratories to lives.

One phrase of a claim

One email a week. The research, explained.

A journal or book title

Published in Molecular Psychiatry; explained in The Folate Fix.

Interface and data, in sans

Search 134 papers · 2006–2026

At most one .voice phrase in a section. On deep surfaces the voice turns aqua.

The label grammar

.kicker.kicker--strong.sep.fmt

Paper·2018·Clinical trial

Folinic acid improves verbal communication in children with autism and language impairment

Frye RE, Slattery J, Delhey L, et al. Molecular Psychiatry, 2018

Above a title

Format · Year · Kind or Format · For families

Below it

Authors, then the journal in italics, then the year.

In a mixed list

Video·For families·4 min

The label is words. An icon goes in front only where a glyph is the fastest way to tell media apart in a mixed list (.fmt). An icon on every label is the mark of a template.

The reading column

.wrap--read.prose.prose--sans.measure

Leucovorin, also called , is a prescription form of folate that has been used in medicine for over 80 years. Dr. Frye has studied it in autism for about two decades.

In 2018 his team ran the first double-blind, placebo-controlled trial. For 12 weeks, 23 children took leucovorin and 25 took a placebo. Verbal communication improved more in the children who took leucovorin.1 Since then, independent groups in 4 countries have run blinded trials of their own.

What it is not

Leucovorin is not an “autism pill”. A clinician weighs three things:

  • the right patient,
  • the right dose,
  • the right formulation.

Reviewed by Richard E. Frye, MD, PhD. This page is for education and is not medical advice.

Serif at on a column of : about 65 characters a line, inside a .wrap--read container. .prose--sans sets the same column in the sans at body size, for text that reads as interface, like the last line here. Links in prose are always underlined.

Numerals and the highlighter

.nummark.hl

Default figures

$2,000,000
$111,111
$67,000

Tabular, .num

$2,000,000
$111,111
$67,000

Use .num wherever figures stack: amounts, counts, tables. Figures are always lining, never old-style.

Leucovorin is not an “autism pill”: it needs the right patient, dose and formulation.

The yellow highlighter marks one key phrase in a passage: <mark>, or .hl on any element. It is the only yellow apart from the caution note.

What not to do

The evidence

Five blinded trials.

Each dot is one child. Body text is 17px or larger.

DoDisplay type is light. Sentence case everywhere; capitals only in the label. Body text at 17px or more, captions at 13.5px or more.

Five blinded trials.

Each dot is one child. This sentence is set at 12px, which is too small to read on a phone held at arm’s length.

Do notNo bold display lines, no capitals outside labels, no body text under 17px.
Space, grid and shape05

A 4px base, a 1280px page, square corners.

Space comes from one scale. Sections, gutters and gaps breathe with the viewport. Layout is a small set of classes that all collapse to one column on a phone.

The spacing scale

Drawn to scale

  • --s1
  • --s2
  • --s3
  • --s4
  • --s5
  • --s6
  • --s7
  • --s8
  • --s9
  • --s10

Fluid measures

Value at this viewport

TokenNowJobDefinition
--sectionVertical padding of .section
--section-tightVertical padding of .section--tight
--gutterThe page margin, inside .wrap
--gapThe gap of every grid
--maxThe container
--measureThe reading column
--header-hThe header; offsets sticky things
--util-hThe utility bar above it

The container

.wrap.wrap--read.measure.section.section--tight

.wrap is 1280px at most with a fluid gutter each side; .wrap--read is 832px for reading pages; .measure caps a text block at 640px. One .section or .section--tight per band, with the surface class on the same element; .section--flush-top removes the top padding when two bands share a surface.

Layout classes

Shown at desktop width; each is one column on a phone

.splitTwo equal columns from 900px. .split--center centres them vertically.
.split--7-5Statement and aside.
.split--5-7Aside first.
.split--8-4Content and a narrow column.
.split--4-8The narrow column first.
.with-railMain column plus a 19rem rail, from 1000px.
.with-rail--leftA 17rem left rail, for filters.
.cols-2Two equal columns from 640px.
.cols-3Two from 640px, three from 960px.
.cols-4Two from 640px, four from 960px.
.stackVertical rhythm; --stack sets the gap.
.clusterInline items that wrap; --cluster sets the gap.
.sec-head--splitThe section head: bar with label and folio, statement, lead.
.stickySticks below the header, as the index beside this page does.

Every grid starts as one column that cannot be pushed wider than its container. .grid is the bare grid with the standard gap, for a page’s own columns.

Radius

Square by default

  • --r-contentCards, photos, figures, tags, notes
  • --r-controlButtons, menu items
  • --r-fieldInputs and selects
  • --r-toolPanels, the giving widget, popovers
  • --r-pillSearch, chips and switches only

Tool·Printable

A one-page sheet for your doctor

DoSquare buttons and cards. A chip is round because it is a filter you switch on and off.
Read the evidencePlain summary

Tool·Printable

A one-page sheet for your doctor

Do notNo pill buttons, no pill tags, no rounded cards, no card shadows. Nothing has a large radius.

Elevation

--shadow-float

Flat

A panel, a card, a row: a hairline, never a shadow.

None for anything that sits in the page.

Floating

Suggestions, a definition, the toast, the topics flyout, the giving widget and the search field on deep.

One shadow for the few things that float above the page:

Utilities

.muted.quiet.center.right.mt-0 to .mt-8.rule-top.rule-strong.balance.pretty.only-wide.only-narrow.vh.needs-js.no-print

.rule-strong a 2px rule above, as on a section bar

.muted secondary text

.quiet meta text

.rule-top a hairline above

.center

.right

A heading with .balance breaks into lines of even length.

A paragraph with .pretty avoids ending on a line of one short word, which matters most in a lead.

.only-wide You are at 900px or wider, so this line shows.

.only-narrow You are below 900px, so this line shows.

.mt-8 puts 64px above this line. .mt-2 to .mt-8 follow the spacing scale; .mt-0 removes a margin.

.vh hides text visually and keeps it for screen readers. .needs-js hides a control when scripts are off. .no-print keeps a thing out of print. Reach for a utility to place a component, never to restyle one.

Motion06

Small, once, and never in the way.

Motion confirms what the reader just did. It stops under reduced motion, and never hides content.

Durations and easing

--dur-1 to --dur-4--ease

  • --dur-1
  • --dur-2
  • --dur-3
  • --dur-4

--dur-1 for colour and border changes; --dur-2 for arrows, chevrons and switches; --dur-3 for the header and the bars that slide in; --dur-4 for image scale and entrances. One easing for all of it: .

The hover vocabulary

Hover or focus each one

An arrow moves 4px

.go, .btn .i--go, .tool__go

.row, .card, .step: the whole block is the target

The top rule thickens2px to 4px

.tool; a .topic line does the same with its bar

.card, .media, over

Entrances

data-rise--i

First

Rises 14px and fades in.

Second

--i:1 delays it 70ms.

Third

--i:2 delays it 140ms.

The only entrance. It plays once, when the element scrolls into view. Without scripts, or under reduced motion, the content is simply there. Charts never wait for a scroll: they render complete on load.

Never used

  • Autoplay, of video or of anything else
  • Parallax and scroll-driven scenes
  • Counters that count up
  • Carousels and sliders
  • Pop-ups and interstitials
  • An empty chart that draws on scroll
Icons07

One stroke, always beside a word.

A 24px grid, a 1.6px stroke, round caps, drawn in the current text colour. An icon never carries meaning alone: it sits beside a label, or its control has an aria-label.

Formats

.i.i--sm.fmt

  • i-paper
  • i-read
  • i-handout
  • i-video
  • i-audio
  • i-course
  • i-book
  • i-ask

Controls and signs

  • i-go
  • i-ext
  • i-down
  • i-search
  • i-menu
  • i-close
  • i-play
  • i-heart
  • i-check
  • i-copy
  • i-print
  • i-mail
  • i-clock
  • i-info
  • i-alert
  • i-lock
  • i-people
  • i-flask
  • i-doctor
  • i-filter
  • i-plus
  • i-minus
  • i-type

VideoPodcastHandoutBuy on Amazon

.i is 1.25em; .i--sm is 1em. .fmt pairs a format glyph with its word. A link that leaves the site carries i-ext and names where it goes.

Controls08

A control says what will happen.

A label starts with a verb and names the result: “Read the evidence”, never “Learn more”. Every target is at least 44px. Focus is always visible: a 3px ring, teal on light surfaces and yellow on deep.

Buttons

.btn.btn--secondary.btn--give.btn--lg.btn--sm.btn--block

VariantDefaultHoverFocus
.btnPrimary. One in a view.Read the evidence Read the evidence Read the evidence
.btn--secondaryA second action.Search 134 papersSearch 134 papersSearch 134 papers
.btn--giveGiving, and nothing else.DonateDonateDonate

Sizes

A block button fills its column

The hover and focus columns are static renderings; every button in the first column is live. A button is pressed 1px down while it is held. On deep the primary button turns light, the give button brightens, and the focus ring turns yellow.

Text actions

.go.go--fg.link

.go is the default way to say “more”: a label and an arrow. A link inside a sentence is underlined, always. A link that leaves the site says so in its label and carries i-ext.

Chips and tags

.chip.chip__n.chip--sm.tag.tag--line.tag--ok.tag--warn.tag--give

Chips are filters you switch on and off. They are round.

Tags are static labels. They are square.

Plain summaryOpen accessFundedPre-orderNot yet funded$104,000 raisedExample

.placeholder-tag marks any value ADTF has not supplied. It stays until the real figure replaces it.

Segmented control

.seg.seg__opt.seg--block.seg--on-paper

The audience switch, working. The page now has data-audience="" on its root, and the choice is remembered.

.seg--block fills its column, as in the giving widget.

.seg--on-paper keeps the pale track when the control sits on a white panel. It is for light surfaces only.

One choice of a few: audience, frequency, reading register. A page shows or reorders content with CSS on the audience attribute. It never colour-codes by audience, and never hides the only route to something behind the switch.

Fields

.field.field__label.input.select.field__hint.field__error.check.check__n

Weekly, on Mondays. We never share your address.

A select keeps its arrow on the right.

Enter an email address like name@example.com.

Topic

A label is always visible, above its field. A hint sits below. An error replaces the hint, says how to fix the problem, and carries an icon so it is not red alone.

Search

form.search[data-search].search--lg.suggest.suggest__group.suggest__opt.chip[data-q]

Try

The suggestions for “leucovorin”, as a static rendering. The first row is the highlighted one.

The largest control on the site, and one of the three round shapes. It searches all 189 items as you type, groups the suggestions by kind, and treats words that mean the same thing as the same: “leucovorin” finds “folinic acid”. Arrow keys move, Enter opens, Escape closes. .search--lg is the taller one in the search dialog.

Tabs

.tabs[role=tablist].tab[role=tab].tab__n

Everything searchable: 189 items.

Arrow keys move between tabs; Home and End jump to the ends. A tab that filters a list shows its count. The strip scrolls sideways on a phone rather than wrapping.

Disclosure

details.acc.acc__body

What is leucovorin?

A prescription form of folate, also called folinic acid. It has been used in medicine for over 80 years.

Is it an “autism pill”?

No. It needs the right patient, dose and formulation, and the decision belongs with your child’s doctor.

Who has tested it?

Dr. Frye led the first blinded trial, published in 2018. Since then, independent groups in 4 countries have run blinded trials of their own.

A native details element, so it works without scripts. Use it to keep a phone page short: secondary detail collapses, the main answer never does.

Pagination and breadcrumbs

.pager.crumbs.toc

A long list pages; it never loads without end. .toc is the in-page index with a marker for the section in view: the rail beside this page is one.

Content components10

Rows for everything, cards for the few.

A paper, a handout, a video and a course are all listed the same way: one row, with one label grammar. Cards are kept for the few items a page chooses to feature. Boxes are kept for tools.

Library rows

.rows.row.row__kicker.row__title.row__dek.row__meta.row__aside.row__out.with-rail--left

Filters, in a .with-rail--left

Papers · <!--#gen rows topic=cerebral-folate-deficiency landmark=1 limit=2 -->

Family resources · <!--#gen resource-rows topic=cerebral-folate-deficiency limit=2 -->

Courses · <!--#gen course-rows limit=2 -->

The whole row is the link. A paper that opens the journal says so on the right; the one paper with a page here is tagged “Plain summary”. Course rows carry length, audience, price and “No CME credit”.

Compact rows and rows with a thumbnail

.row--compact.row--thumb.row__thumb.fmt.with-rail

For media, in a mixed list

In a .with-rail · compact=1

Feature cards

.card.card--lead.card__media.card__title

At most three in a row, and only for chosen items. One of them may lead. A card has no box, no shadow and no icon: an image at 3:2, the label, a serif title.

Tool card and panel

.tool.tool__title.tool__foot.tool__go.panel

Tool

Build your doctor sheet

Print the sheet

A tool is a task, not an article, so it gets a box: a strong rule on top, a serif title, an arrow. A panel is the ruled container the tool itself sits in.

Topic lines

.topics.topic.topic__q.topic__name.topic__n--share

<!--#gen topics limit=3 -->

Could my child's brain be low in folate?

Cerebral folate deficiency · When folate cannot reach the brain, and the treatment that gets it there.

11papers

Why does my child lose skills when they are sick?

Mitochondrial dysfunction · How cells make energy, and what happens when they cannot keep up.

48papers

Which treatments have been tested properly?

Treatments · Trials and studies of specific treatments, and how strong the evidence is.

46papers

The contents page of the site. The parent’s question leads; the scientific name follows. The rule under each line is a bar: its length is the topic’s share of papers, measured against the largest topic.

In brief, notes and the caution

.brief.brief__title.note.note__title.note--caution

In brief

  1. Leucovorin is a prescription form of folate.
  2. 5 blinded, controlled trials have tested it in 293 children.
  3. It is not an “autism pill”.

A caution

Leucovorin is not an “autism pill”. It needs the right patient, dose and formulation.

Not medical advice

This page is for education. Decisions about treatment belong with your child’s doctor.

Three numbered points come before a long page, and first on a phone. A note always has a title and an icon, never colour alone. The caution comes before or beside any number about a treatment, never screens later.

Reviewed stamp, quote and citation

.stamp.stamp__who.quote.quote__by.cite.cite__text.cite__row.doi[data-copy]

Reviewed by Richard E. Frye, MD, PhDLast reviewed November 2026Example

Moving treatments from laboratories to lives.

The foundation’s brand line

Frye RE, Slattery J, Delhey L, Furgerson B, Strickland T, Tippett M, et al. Folinic acid improves verbal communication in children with autism and language impairment: a randomized double-blind placebo-controlled trial. Molecular Psychiatry. 2018;23(2):247–256.

doi:10.1038/mp.2016.168

A quote is set in the first voice at statement size, and is only ever something a named person or document said. A citation is given in full, with its DOI and a button that copies it.

People

.person.person--lg.person__img.person__name.person__role

Leadership · <!--#gen people group=leadership -->

Richard E. Frye, MD, PhD

Chief Scientific Officer · Cofounder

Daniel A. Rossignol, MD, FAAFP

Cofounder

Adrienne C. Scheck, PhD

Research Scientist

Kelly Sirju

Chief Operating Officer

Stella V. Laurella, CPA, MBA

Treasurer

Research team · <!--#gen people group=team -->

Hamsa Raman, MA, CCC-SLP

Clinical Research Coordinator

Sarah Lychy, ND

Investigator

Zoë Hill

Genetic Specialist

Jaclyn Bain, ND, FABNP

Investigator

Christy Coleman, RN, BSN

Pediatric Clinical and Research Specialty Nurse

Cara McCormack-Frye, BS

Pediatric Clinical and Research Associate

Sumedh Seetharam

Research Assistant

Every portrait is greyscale and cropped to one shape, by rule. That makes uneven head shots look intended.

Steps

.steps.step.step__title

Diagnostics

Find a measurable difference that can be treated.

Research

Study how it works in the laboratory.

Clinical practice

Test a treatment in the clinic, in blinded trials.

Treatments

Make what works practical and safe for families.

For a sequence the reader follows in order. Put a link in the title and the whole step becomes the target.

Chapters

.chapters.chapters__rail.chapter.chapter__year

2013

In 93 children with autism, about three in four tested positive for folate receptor autoantibodies. It was an open-label study with a waitlist comparison group.

2018

The first double-blind, placebo-controlled trial: 23 children on leucovorin, 25 on placebo, 12 weeks. Verbal communication improved more with leucovorin.

2025

In September the FDA approved a label change for leucovorin for cerebral folate deficiency. The foundation published its response on 1 October.

A story told beside a rail that stays in view and marks the chapter you are reading. The rail appears from 960px; on a phone the chapters simply follow one another.

Media and photographs

.media.media__play.media__tag.photo.photo--mono

A boy at a kitchen counter, in a still from CBS Evening News CBS Evening News Play video · 4 min

A still, a label that says what plays and for how long, and an optional tag. Video never loads before a click.

Dr. Richard Frye in his laboratory

.photo crops an image to its box; .photo--mono makes it greyscale, the treatment for every portrait.

Glossary term, and plain or science

.term[data-def].term-pop[data-register][data-reg].seg[data-register-switch]

In the first blinded trial, children who took leucovorin for 12 weeks improved more in speech than children who took a placebo. Leucovorin is not an “autism pill”.

In a 12-week , placebo-controlled trial of folinic acid (23 treated, 25 placebo), verbal communication improved relative to placebo. Patient selection, dose and formulation matter.

The definition a term opens, as a static rendering

Double-blindNeither the families nor the researchers knew which children received the medicine until the trial ended.

The same finding in two registers. Plain is the default, the choice is remembered, and with no script the plain version shows. A term with a dotted underline opens its definition: press “double-blind” in the science version, or “folinic acid” in the reading column above. Explain a term the first time it appears.

Empty state

.empty

Nothing for “xylophone” yet.

Try a topic such as folate, mitochondria, seizures or gut.

See all 134 papers

Say what happened, then offer the next step. Never a blank area.

Data11

One mark, one thing.

Each dot is one real paper, one child in a trial, one project. Every figure is captioned, sourced, and stated in text for people who cannot see it.

Unit marks and the legend

.units.u.u--off.u--dim.u--hl.u--sq.legend--u--u-gap

.u one counted thing

.u--off the comparison group

.u--dim not in this subset

.u--hl the one to notice

.u--sq things, not people

--u:7px smaller, for a phone

Assigned leucovorinAssigned placebo

A unit chart always has role="img", an aria-label that states the numbers, and a legend. Marks smaller than 24px are never links: they are too small to tap.

The record

.record.record__col.record__yr.record__read

134 papers, 2006–2026. One mark is one paper, stacked by the year it was published.Source: the ADTF library, counted 10 November 2026.
11 of 134 papers are on cerebral folate deficiency. The rest are dimmed.Source: the ADTF library, counted 10 November 2026.

<!--#gen record --> draws it from the data at build time; hl=topic-slug lights one topic. With a mouse, pointing at a mark names its paper in the line beneath. Put the generator in its own div.

A figure: Fig. 1, the trials

.fig.fig__cap.fig__label.fig__src.trials.trial

Assigned leucovorin, 155Assigned placebo, 138

Fig. 1The 5 blinded, controlled trials of leucovorin in autism. One mark is one child. The bracket under each group is the length of the trial.Source: ADTF statement, 1 October 2025. Countries are taken from the authors’ affiliations.

<!--#gen trials -->. Every chart sits in a .fig with a label, a caption that says what one mark is, and a source. Design, sample size and comparison appear with every trial, before any result.

Table

.table-scroll.tabletd.num

The blinded, controlled trials of leucovorin in autism
YearTrialJournalCountryDesignLeucovorinPlaceboWeeks
2018Frye et al.Molecular PsychiatryUnited StatesDouble-blind, placebo-controlled232512
2020Renard et al.BiochimieFrancePlacebo-controlled, randomized91012
2021Batebi et al.Child Psychiatry & Human DevelopmentIranDouble-blind, placebo-controlled333310
2024Panda et al.European Journal of PediatricsIndiaDouble-blind, placebo-controlled404024
2025Zhang et al.NutrientsChinaBlinded, placebo-controlled503012

The text equivalent of Fig. 1. A table always sits in .table-scroll, so on a phone it scrolls inside itself and the page never does; give that wrapper tabindex="0" and a label so a keyboard can scroll it. Figures are right-aligned and tabular.

Stats

.stats.stat.stat__n.stat__label.stat__src

293children in 5 blinded, controlled trialsADTF statement, 1 October 2025
134scientific papers, 2006–2026ADTF library, 10 November 2026
DoA number sits beside the evidence it summarises and carries its source or date. Never more than three together.
134papers
16projects
9courses
Do notNo full-width band of counters, no number without a source, and no percentage before its caveat, its comparison and its study design.

Meter

progress.meter

$104,000 raised of $260,000Example

Progress toward a stated goal, with the amounts in text beside it. Only real progress is a bar: a goal drawn as a bar reads as “100% funded”. Until ADTF supplies amounts raised, every one shown is tagged as an example.

The plan

.plan.plan__program.plan__blocks.plan__block.plan__label

The $6.05 million plan, to scale. One block is one of 16 projects, as wide as its goal, grouped into 4 programs. It shows size, never progress.Source: ADTF project list.
Light a project

<!--#gen plan -->. Adding .is-on to a block lights the project a donor has chosen; it also follows the giving widget in section 12. This and the meter are the only places orange appears in data.

Giving12

The widget is the strongest object on its page.

Step one asks three things only: how often, how much, for what. Four rising amounts with the second chosen, plus “other”. The button states the gift. Inside the widget only that button is orange.

The giving widget

form.give[data-give].give__amounts.give__amt.give__other.give__outcome.give__dest.give__assure.give--light

Secure checkout on the next stepTax-deductible · 501(c)(3)

On a light surface the widget is an ink panel.

Goes to

Secure checkout on the next stepTax-deductible · 501(c)(3)

On deep it is a white panel that floats.

.give--light is the same white panel without a deep section around it. This one is cut down to the three questions.

All three are working: nothing is sent, and the button reports what it would do. The widget is never a tinted band. On a phone it comes before any list of projects. There is no line that prices an outcome (“$50 pays for…”): ADTF has not supplied unit costs. A button with data-fund="2B" anywhere on the page aims the first widget at that project and scrolls to it; so does ?project=2B in the address.

Project rows, in three states

.program.program__head.project.project__code.project__title.project__money.project__act[data-fund]

<!--#gen projects program=4 -->

Program 4 · $400,000

Advanced laboratory equipment

The instruments that make the measurements possible.

4A

Seahorse XF Pro Analyzer

Goal $260,000

4B

Cytation 5 Cell Imaging Reader

Goal $67,000

4C

CFX Opus 96 Real-Time PCR System

Goal $35,000

4D

LI-COR Odyssey XF Imager

Goal $38,000

Part funded, with a meter and its example tag; funded, with a tag and no button; and not yet funded, with a goal and no bar. “Fund 4A” aims the widget above.

Trust line and funders

.trust.logos

A 501(c)(3) non-profit foundation Secure checkout EIN 00-0000000To confirm

Research supported in the past by

The BRAIN Foundation The O’Sullivan Foundation US–Israel Binational Science Foundation The Jonty Foundation XEL Foundation Department of Defense

Funders’ logos are greyscale and sit on a light surface only. No charity-rating badge appears until ADTF holds the rating.

The phone give bar

.give-bar[data-show]

On a page where giving is the point, a bar slides up from the bottom of a phone once the page’s own widget has scrolled out of view, and leaves when it returns. It links to the widget, not to the top of the page. It never appears from 900px up.

Support the research.
Tax-deductible · 501(c)(3)
Donate
Books and courses13

Buying happens elsewhere, and says so.

There is no cart and no price in the header. A buying button is ink, never orange, names the retailer and says you are leaving. Books and courses are sold by Metabolic Learning Resource, founded by Dr. Frye: say so once, plainly, near the button.

Cover, facts and hand-off rows

.coverdl.facts.buy

Front cover of The Folate Fix, by Richard E. Frye

Book·2025·For families and clinicians

The Folate Fix

Exploring the role of folate in autism and neurodevelopmental disorders. By Richard E. Frye, MD, PhD.

Format
Paperback
Pages
332
Published
2025
CME credit
None

Sold by Metabolic Learning Resource, founded by Dr. Frye. These links leave this site.

A cover is shown flat, at one size, on a neutral tile: no 3D mock-up, no badge, no price on a card. Never write “proceeds fund research”. Say “No CME credit” where a clinician would look for it.

Syllabus

ol.syllabus

Series·Mitochondrial medicine·5 courses

  1. The Basics of Mitochondrial Function9 chapters
  2. Mitochondrial Dysfunction and Autism4 chapters
  3. The Role of the Mitochondria in Brain Development and Function12 chapters · 47 min
  4. Diagnosing Mitochondrial DisordersPre-order
  5. Treatment of Mitochondrial DysfunctionPre-order

A numbered list of what a course or a series contains, with a length or a status on the right. The index at the top of this page is one too.

The enrol bar

.enrol-bar

On a course page the price and the way to enrol stay pinned to the bottom of a phone. The button is ink: buying is not giving. It names Thinkific, where the course is taken, and carries the leaving-the-site arrow.

Rules14

What every page passes, and what was turned down.

A short list to check a page against before it ships. The full reasoning is in the written specification, design-system.md.

Accessibility: WCAG 2.2 AA

  • One h1. Headings step down without skipping.
  • Landmarks, and a visible label on every control.
  • Focus is visible and is never removed.
  • Targets are at least 44px.
  • No meaning by colour alone: a word or an icon beside it.
  • Alt text on meaningful images; empty alt on decorative ones.
  • Every chart states its numbers in text.
  • Everything works from the keyboard, and with scripts off.
  • Everything stops under reduced motion.
  • No sideways scroll from 360px to 1920px. A phone page is about 16 screens at most.
  • No libraries and no web requests. Images are sized and lazy.

Content: true, plain, sourced

  • Only facts from the fact sheet and the data. Nothing invented.
  • Every number has a source or a date. Counts are generated, never typed.
  • A placeholder is visibly tagged.
  • No response rate without its comparison and its study design.
  • The caution (“not an ‘autism pill’”) comes before or beside any efficacy number.
  • Lead with the parent’s question. The scientific term comes second, explained once.
  • Sentences under 25 words in copy for families.
  • Button labels start with a verb. A link that leaves the site says so.
  • Tell each story once on a page.
  • No promise of outcomes. Neutral on politics.
  • Sentence case everywhere; capitals only in labels.

Rejected: do not bring these back

Turned down by the owner, or killed in user testing

  • A four-colour brand stripe
  • Colour-coded audience cards; persona pills
  • Brown-black bands; dark bands beyond the opening and one chapter
  • Heavy or bold display type; thin giant numerals
  • Roboto
  • A hero photo on an offset colour block
  • Puzzle imagery
  • Pastel tints: green, yellow or peach sections
  • Rounded cards and pill buttons
  • Colour head shots
  • A row of stat counters; a masthead strip
  • Keyboard-hint badges
  • A goal drawn as a progress bar
  • Dots used as links
  • An icon in front of every label
  • Journal furniture at 12px (“n = 138”, “Colophon”)
  • Anything that reads as a template

This page checks itself

Read from the live stylesheet

    If a class or a token is added to system.css and not shown here, this list says which.

    ADTF Weekly

    One email a week. The research, explained.

    A short video from Dr. Frye on one new paper, what it means for families, and what the foundation is working on. Free, and you can leave at any time.