Questions, answered.

GeneFox is a native app for exploring public gene-expression and cancer-genomics data on a phone, tablet, or desktop. You pick a gene and a cohort, and it draws the figure — expression by tumour stage or subtype, tumour versus normal, a survival curve, a clustered heatmap — and reports the statistics behind it.

It is a reading and exploration tool, not a pipeline. Nothing needs to be installed, configured, or scripted, and the figures are computed on the device in front of you rather than on a server we operate.

Scientists who work with gene-expression data and want to look something up without writing code. If you have ever wanted to check whether a gene is higher in tumour than normal, or whether its expression tracks with survival, and did not want that to become an afternoon of R, GeneFox is built for you.

It is equally useful to computational biologists as a quick look-up, but no command line, R, or Python is required at any point.

GeneFox is a genuinely native app on each platform rather than a wrapped web page, so the charts, gestures, and system integration behave the way they should on the device you are holding.

An internet connection is needed to fetch a dataset, because the underlying data lives in public research repositories. Once a dataset is loaded, the analysis itself runs locally.

From established public repositories, requested directly by your device:

• NIH Genomic Data Commons (GDC) — TCGA and related programmes; 65 supported cohorts covering the major tumour types.
• UCSC Xena — harmonised expression, phenotype, and survival data for those cohorts.
• NCBI Gene Expression Omnibus (GEO) — over 40,000 historical availability predictions used as search hints. These are not verified results; GeneFox verifies a Series when you open it.
• ARCHS4 — a raw-count fallback for GEO-classified human or mouse expression RNA-seq Series after a directly usable Series Matrix and NCBI-generated counts are unavailable. GeneFox accepts only complete public-GSM coverage, never silently truncates a cohort, and does not call these values TPM.
• GTEx — healthy-donor expression served through Xena’s TOIL compendium for scale-compatible exploratory comparison. GTEx remains an unpaired cross-study reference, without adjustment for study or collection effects.
• CCLE via Treehouse — 933 cancer cell lines, with six fixed metadata fields: cell-line name plus five grouping dimensions — TCGA Acronym, Tissue, disease, Histology, and Subtype/subcategory.
• Treehouse Tumor Compendium 25.01 PolyA — 1,761 pediatric and rare-tumour samples across 25 disease cohorts, drawn only from contributing studies independent of TCGA and TARGET: St. Jude Cloud, the Sequence Read Archive, EGA, ICGC, and the Children’s Brain Tumor Network. The compendium’s own TCGA and TARGET specimens are excluded, because they are the same specimens GeneFox already serves from the GDC hub, re-quantified through a different pipeline.

GeneFox does not host or re-publish these datasets. It reads them from the providers and shows you the result, and each provider’s own site remains the authoritative source.

• Expression by group — stratify a cohort by stage, subtype, mutation status, or other clinical variables, as box, bar, or strip plots; Android also offers violin plots.
• Tumour versus normal — participant-level exploratory comparison with adjacent-normal samples and, where supported, a tissue-matched GTEx healthy-donor reference; GTEx is unpaired and cross-study.
• Pan-cancer — one gene across supported, analyzable TCGA cohorts.
• Pediatric and rare tumours — one gene across the independent-source Treehouse cohorts, grouped by disease, pediatric/AYA status, sex, or source study, with CSV export.
• Copy number and methylation — alongside RNA-seq expression, you can browse gene-level GISTIC2 copy number and Illumina 450K methylation for the supported TCGA cohorts.
• Survival — Kaplan–Meier curves for overall, disease-specific, progression-free, and disease-free endpoints, with median survival and eligible two-group log-rank tests. Eligible High-versus-Low views also present a Cox hazard ratio with 95% confidence interval and a proportional-hazards diagnostic or a clear non-estimable explanation.
• Differential expression — volcano plots with multiple-testing control.
• Heatmaps — gene panels with hierarchical clustering and annotation tracks.
• Correlation — Pearson correlation between two genes within a cohort.
• Signature scoring — 41 documented hallmark, immune, oncogenic, and pathway presets, or your own list of genes.

The tests are named on screen and carried into every export, so a figure is never a black box:

• Mann–Whitney U for eligible two-group expression comparisons, and Kruskal–Wallis when a split has more than two eligible groups.
• Eligible two-group survival comparisons use a log-rank test with exact conditional label permutation when tractable and an asymptotic chi-square tail otherwise; median survival is reported when reached. Eligible High-versus-Low expression views and Android’s signature-score survival panel separately fit a univariable Cox proportional-hazards model and present its hazard ratio with 95% confidence interval, proportional-hazards diagnostic or warning, and a reason when inference is not estimable. The signed log-rank direction statistic remains distinct from that Cox hazard ratio. Pan-cancer survival applies Benjamini–Hochberg false-discovery-rate control across analyzable cohorts.
• Benjamini–Hochberg false-discovery-rate control for differential expression.
• Pearson r correlation.

Every statistic and figure is computed on your device.

Yes. You can import your own expression matrix and analyse it with the same tools used for the public cohorts.

The imported copy remains in app-private storage and is not uploaded by GeneFox or automatically attached to Cloud Assist. Apple may include app-private data in user-controlled iCloud, Finder, or device-transfer backups that Vindhya Data Science cannot access; operating-system backup is not developer collection. Public Android v1 disables app backup and device transfer. Anything typed in a question that is sent to Cloud Assist is transmitted to Google.

Yes. Apple devices export reports as PDF or HTML; Android exports reports as HTML. Underlying numbers can be exported as CSV. Reports carry an automatically generated methods section naming the tests, data source, and sample counts behind each figure.

No GeneFox account, login, or profile is required. The optional updates request unlocks nothing.

• No ads, no third-party tracking, no behavioural analytics.
• Statistics and figures are computed on your device. Scientific requests go to the public providers you choose; supported production iPhone, iPad, native macOS, iPad-on-Mac, and Android execution can also use the disclosed optional Cloud Assist path. Supported Apple destinations initialize the narrow Firebase configuration path at launch; Android leaves Firebase uninitialized until a consent-authorized Cloud Assist request.
• Bookmarks and saved cohorts stay on your device and, on Apple devices, sync through your own private iCloud account — we cannot see them.
• Apple may include app-private imports, workspaces, reports, and settings in user-controlled iCloud, Finder, or device-transfer backups. Vindhya Data Science cannot access those operating-system-managed copies, and they are not developer collection. Public Android v1 disables app backup and device transfer.

The full detail is in the Privacy Policy.

An optional feature on supported production iPhone, iPad, native macOS, iPad-on-Mac, and Android execution for questions the local interpreter cannot handle. An unsupported or misconfigured destination fails closed to local interpretation.

• No cloud request is sent without a one-time authorization or previously granted standing consent.
• It sends the typed question, interpretation instructions, capability flags/tokens, and response-schema metadata to Google Gemini through Firebase AI Logic. It never automatically attaches data files, dataset titles or identifiers, expression values, bookmarks, or chat history.
• Production requests use App Attest on supported physical iPhone/iPad hardware, DeviceCheck on supported native macOS and iPad-on-Mac execution, and Play Integrity on Android.
• Cloud Assist is limited to adults 18 or older using it for professional or business scientific research. That restriction applies to Cloud Assist alone; the rest of GeneFox has no age requirement.
• Do not type personal, patient, or confidential information into it.

Everything else in GeneFox works whether or not you ever enable it.

No. GeneFox is an independent application developed by Vindhya Data Science. It is not a government app, and it does not represent, and is not affiliated with or endorsed by, any government entity or agency (including the U.S. National Institutes of Health, National Cancer Institute, or NCBI) or any of the data providers it reads from.

Consult each provider’s official site for the authoritative source.

Analyses produced by the app are exploratory. Survival curves, differential-expression results, and natural-language answers should be validated against the source data and the primary literature before you draw a scientific conclusion, and its outputs have not been independently validated for clinical decision-making.

The “GeneFox updates” signup is entirely optional and unlocks nothing — every feature works the same whether or not you sign up.

If you did sign up, you can unsubscribe in the app’s Settings with a single tap, or email info@vindhyadatascience.com and ask us to remove your address.

Email info@vindhyadatascience.com. If you are reporting something that looks wrong in a figure, telling us the gene, the cohort, and the analysis makes it much faster to track down.