Detect cryptic splicing from any knockdown or perturbation, design and validate the assays that confirm it, and keep the whole workflow — from primer to gel to lab notebook — in one place. Eighteen tools, one app, entirely on your own machine.
A real Mac app — not a browser tab · macOS 12 or later · Apple silicon & Intel · installs R for you on first launch
Every number Lit2Bench shows you comes from plain arithmetic or a real external tool — primer3, UCSC, NCBI — never from a guess. That is a design constraint, not an accident.
Primers come out of primer3. Sequence and annotation come from the
UCSC REST API, citations from NCBI E-utilities. Nothing is approximated when the
authoritative source is one request away.
Everything runs on your own Mac — there is no account, no backend and nothing to sign into. Multi-gigabyte BAMs are read off your disk in place, never uploaded. Notebook entries are plain JSON files you own.
A plain-language interpretation step can run against a local Ollama model. It is off by default, never leaves your machine, and is only ever handed the numbers the detector already computed.
Pick an exon in one tool and it arrives pre-filled in the next. Filter by what you're doing, or search for the thing you need.
Reads control vs. knockdown BAMs over a locus and screens for all four recognised splicing defects — cryptic exon inclusion, cryptic splice-site selection, exitrons and intron retention — in an IGV-style sashimi plot. Double-click to zoom.
DesignDesigns a junction-spanning primer pair against live reference sequence and shows the expected canonical vs. cryptic-exon product sizes, with a schematic you can export.
DesignReusable procedures and the experiments you spin up from them, with editable tables for setup and results. Saves to disk as JSON and reopens across sessions.
NotebookRuns the full cryptic-detection pipeline across a whole gene or locus list against one BAM pair, returning a row per locus. Click any hit to open it in the engine.
DesignSearch a gene symbol, RefSeq/Ensembl transcript ID or a locus and see every annotated isoform in the region — strand, length, coding status, exon and intron counts, CDS and UTR spans.
DesignPulls real exon and intron sequence for a chosen transcript, exports BED/FASTA/CSV/JSON/GTF, and hands a chosen exon straight to the primer designer — no copying coordinates by hand.
DesignQC Plasmidsaurus (or any) reads against a reference plasmid: local pairwise alignment on both strands, a PASS / GENE_FOUND / FLAGGED verdict, called substitutions and indels, and NCBI screening of unmatched flanks.
DesignDesigns the primer pairs to Gibson/NEBuilder-assemble an ordered set of fragments: a gene-specific annealing region sized to your target Tm, plus the homology tail that builds each junction.
DesignJoins your parts end to end, circularises them, and draws the resulting map.
DesignScales stock and final concentrations into a master mix for N reactions plus excess, in the units you actually pipette in.
DesignRelative expression by 2−ΔΔCt against a chosen calibrator sample, with QC warnings when replicates or reference genes misbehave.
AnalysisNormalises target band intensity to a loading control across lanes.
AnalysisFits a standard curve — linear or quadratic — and back-calculates unknown concentrations from it.
AnalysisCollects the session into three paste-ready artifacts: a primer ordering sheet, a templated Methods paragraph and a references list — built from what you actually ran, with blanks marked rather than invented.
AnalysisWorks out lysate, water and dye volumes for equal-protein-mass loading.
CalculatorsConverts A280 absorbance to concentration via Beer–Lambert, using your extinction coefficient and molecular weight.
CalculatorsComputes molecular weight, pI and extinction coefficient directly from an amino-acid sequence.
CalculatorsSolves C₁V₁ = C₂V₂ for stock volume and diluent volume.
CalculatorsNo tool matches that. Try a shorter search.
The tools are built around one loop. Each step hands its result to the next, so nothing gets retyped and nothing gets transcribed wrong.
Point the engine at a locus and a control/knockdown BAM pair. It flags junctions and exons that appear on knockdown but aren't in the reference annotation.
Click a candidate exon and go straight to primer design against live reference sequence — with the canonical and cryptic product sizes you'd expect on a gel.
Scale the master mix, run the reaction, then quantify: ΔΔCt for qPCR, densitometry for the Western, a standard curve where you need one.
Write the experiment up against a reusable procedure, and let the session assemble its own ordering sheet, references and Methods paragraph.
A real run against a real control/knockdown BAM pair — UNC13A and TDP-43 here, but the engine takes any locus and any perturbation. Every number on screen is computed live, and the detection timing is the app's own: nothing is sped up to flatter it.
Every number on screen is computed live — junction counts from the BAMs,
primers from primer3, coordinates and sequence from UCSC.
Detection runs in under ten seconds against two ~9 GB BAMs, because it
reads one locus through the index rather than the whole file, and the clip
holds that wait for exactly as long as it really took.
The download is small because it ships the app, not a runtime. On first launch it finds R — or installs it from CRAN for you — pulls down the R packages, and opens the toolkit in its own window. Every launch after that is instant.
The usual macOS install — nothing unusual in the disk image.
A progress view walks through finding R, installing the packages, and starting up, then hands over to the toolkit. First run takes a few minutes; after that, seconds.
Its own window, its own Dock icon, a normal menu bar, ⌘Q to quit. Exports
land in Downloads; your notebook is stored under
~/Library/Application Support/Lit2Bench.
Expected, and it doesn't mean anything is wrong with the download. Lit2Bench is signed ad hoc rather than with a paid Apple Developer ID, so Gatekeeper won't open it on the first try.
Open System Settings → Privacy & Security, scroll to the bottom, and click Open Anyway next to the Lit2Bench message. You'll only do this once. If you'd rather clear the quarantine flag from a terminal:
xattr -dr com.apple.quarantine /Applications/Lit2Bench.app
Lit2Bench is a script-sourced Shiny app — no package structure and no build step. If you already have R, this is all it takes.
git clone https://github.com/codyciabarra/lit2bench.git
cd lit2bench
Rscript setup.R # installs everything, once
Rscript run.R # launches, opens your browser
setup.R pulls the core CRAN packages plus the per-tool ones
(Rsamtools, GenomicAlignments, pwalign,
httr), so every tool works out of the box.
Two things stay optional: primer3 for primer design
(brew install primer3) and a local
Ollama model for the
interpretation step. The app tells you how to get them if you use a tool that
wants one.
Lit2Bench is built and maintained alongside real bench work, so the roadmap follows what the science actually needs next.
A real double-click install that bootstraps R and every package for you, so using the toolkit doesn't require knowing what CRAN is.
The same bootstrap-on-first-launch approach, packaged as an installer and a desktop launcher. The app itself is already cross-platform — only the wrapper is macOS-specific today.
An Apple Developer ID signature so the download opens on the first click, with no Privacy & Security detour.
The About tab already knows how to update a source checkout. The installed app should check for a new release and update itself the same way.
The Panel Runner covers a gene list today. The natural next step is running detection across the whole annotation and ranking what comes back.
Notebook documents are already portable JSON. Exporting an experiment — figures, primers, methods and all — as one file a labmate can open is the obvious follow-on.
Lit2Bench is built at Stanford University, in a lab studying the genetics of neurodegeneration — including the cryptic-splicing biology the toolkit targets.
Cody CiabarraResearch Intern · Programmer
Aaron D. Gitler, Ph.D.Professor of Genetics · Lab supervisor
Yi Zeng, Ph.D.Postdoctoral Fellow · Code mentor
The Plasmid QC tool is an R port of GeneAlign by Alex Luu. Reference sequence and annotation come from the UCSC Genome Browser; citations from PubMed; primer design from primer3.