# Deep Genomics Financial Model AI-driven RNA therapy platform that uses machine-learning predictors to discover and design genetic medicines for rare and complex diseases. - Canonical: https://finamodel.com/startups/deep-genomics - Excel download: https://finamodel.com/startup-models/deep-genomics.xlsx - Category: Biotech/Pharma - Model type: Biotech rNPV - Funding round: Series C - Funding: $180M - Founded: 2021 - Geography: Toronto (AI platform + preclinical research), Boston (clinical + business development) [DECK, slide 5]. - Customer: B2B ## About the company Deep Genomics uses a Digital AI Workbench of RNA-focused machine-learning predictors to design therapies before costly wet-lab experiments. Its steric-blocking oligonucleotide platform predicts protein restoration, expression increase, and knockdown outcomes, aiming to improve preclinical probability of success from 10% to 50%. At the time of the raise, the company had generated one billion predictions, assessed 300,000 variants, tested 20,000 RNA therapies, and collected 600,000 efficacy and safety data points. The model supports both an internal therapeutic pipeline and partnered assets, including four early discovery programmes with BioMarin. Deep Genomics reported ten programmes, nine first-in-class, and 40 predictors expanding toward 100, but no revenue or deal economics. The model should forecast discovery, preclinical, and clinical milestones; partnership upfronts, milestones, and royalties; internal R&D; probability-weighted programme value; and runway. ## What's included - 5-year monthly revenue build with stage-appropriate growth assumptions - Full P&L, headcount plan, and operating-expense schedule - Cash-flow statement, runway, and burn-rate tracking - Valuation via exit multiple with a DCF cross-check - Returns analysis with MOIC and IRR - Unit economics including CAC, LTV, payback, and cohort retention ## Product & value proposition - Core platform: "Digital AI Workbench" - 40 machine-learning predictors (target: 100 post-Series C) that predict RNA therapy outcomes at scale before costly wet-lab experiments. - Modality: Steric Blocking Oligonucleotides (SBOs) targeting pre-mRNA splicing; three SBO predictor classes: protein restoration, protein expression increase, protein knockdown. - Platform scale at time of raise: 1 billion predictions generated; 300,000 variants assessed; 200,000,000 RNA therapies in silico; 600,000 efficacy/safety datapoints; 250 genes covered; 20,000 RNA therapies experimentally tested. - Key advantage over traditional drug discovery: predictive de-risking of all biology steps (variant → mechanism → therapy mechanism → in vitro effect → off-target safety) upfront rather than sequential experimental trial-and-error. Claims preclinical probability of success improvement from 10% → 50%. - Plug-and-play SBO chemistry enables rapid transfer across targets/indications. ## Market - No explicit TAM/SAM/SOM slide in deck. - Addressable disease spectrum stated as Mendelian (recessive + dominant) through complex large-effect and complex many-effects genetic diseases. - Pipeline estimated worldwide peak sales (proxy for market sizing): - 2019 platform (2 programs, 1 RNA mechanism): Est. $400M peak sales. - 2021 platform (10 programs, 7 RNA mechanisms): Est. >$5B peak sales. - Post-Series C target (4 programs in clinic): Est. $4.5B peak sales. - Individual program estimates: Frontotemporal Dementia $600M–$1B; Niemann-Pick Type C $900M–$1.2B; Wilson Disease $700M–$1.1B; Refractory Gout $800M–$1.2B. - No market growth rate cited. ## Revenue model - No explicit revenue model or pricing slide in deck. - Implied dual-track model: 1. Internal pipeline: develop proprietary programs to clinical stage, then partner/outlicense or take to market. 2. Partnerships: active deal with BioMarin (4 undisclosed programs, all in discovery stage as of 2021). Partnership structure terms (upfront, milestones, royalties) not disclosed in deck. - Slide 14 mentions "access non-dilutive funding" as a partnership benefit, implying milestone/upfront payments expected. - No product revenue, royalties, or partnership economics quantified in deck. ## Traction & metrics - Program count growth: 2 programs (2019) → 10 programs, 9 first-in-class (2021). - Estimated peak sales growth: $400M (2019) → >$5B (2021). - RNA mechanisms: 1 (2019) → 7 (2021). - ML predictors: 40 existing, expanding to 100 post-raise. - Data generated: 600,000 efficacy/safety datapoints; 250 genes; 20,000 RNA therapies tested. - Targets with patented leads: 80. - Genes screened: 100 target (60 complex, 40 Mendelian). - Preclinical programs: 28 (internal + partners + CRO). - Programs in clinic: 4 (est. $4.5B combined peak sales); target 4 in clinic by 2023. - Portfolio projection (from bar chart, slide 13): - 2020: ~6 discovery, ~1 preclinical, 0 clinical - 2021: ~9 discovery, ~1 preclinical, 0 clinical - 2022 (projected): ~15 discovery, ~5 preclinical, 0 clinical - 2023 (projected): ~16 discovery, ~8 preclinical, ~4 clinical - 2024 (projected): ~17 discovery, ~11 preclinical, ~5 clinical - BioMarin partnership: 4 programs partnered, all in discovery/early target validation stage. - No revenue, cash burn, or financial P&L figures disclosed. ## Unit economics - Platform productivity claim: preclinical PoS 10% (industry) → 50% (DG), implying significant cost-per-success reduction, but no dollar figures given. ## Competition / moat - No explicit competitive landscape slide. - Implied moat: - Proprietary AI Workbench with 40 ML predictors across RNA biology (growing to 100). - Causal prediction capability (not just correlation) stated as differentiator. - 600,000+ proprietary experimental datapoints as training set - creates compounding data advantage. - 80 targets with patented leads. - "Plug-and-play" SBO chemistry lowers execution risk per program. - Positive feedback loop: more programs → more data → better predictors → more programs. - Competitive context (RNA therapy space): not named in deck, but landscape includes Ionis, Sarepta, Alnylam, Entrada - none named. ## Team & funding ask / use of funds - Founder & CEO: Brendan Frey PhD FRSC - ML pioneer (University of Toronto, Vector Institute, Microsoft Research, CIFAR). - CBO: Amanda Kay PhD - ex-Genzyme, Pfizer, Harvard, Synlogic. - CMO: Ferdinand Massari MD - ex-Shire, Merck, Pharmacia, Pfizer. - Head of Finance & Bus. Ops: Matt Cahill MBA JD PhD - University of Toronto, Cambridge. - Head of Preclinical Research: Jeffrey Brown PhD - ex-Voyager, WAVE, BMS, Alexion. - Notable advisors: Steve Jurvetson (Future Ventures), Yann LeCun (Facebook/NYU), Vinod Khosla, Jennifer Cook (ex-Roche/Genentech/GRAIL), Peter Barton Hutt (ex-FDA). - Board: Adam D'Augelli (True Ventures), Alex Morgan (Khosla Ventures), Tom Hughes (Navitor), Maryanna Saenko (Future Ventures). - Funding ask: $180M Series C. - Use of funds - stated goals post-raise: 1. Expand predictors from 40 → 100 (complex disease, SBO effect, SBO safety). 2. Screen 100 genes (60 complex, 40 Mendelian); enabled by AI and robotics. 3. Reach 80 targets with patented leads. 4. Expand partnerships (non-dilutive funding). 5. Advance to 28 preclinical programs and 4 programs in the clinic. --- ## Recommended financial model - **Archetype + why:** Biotech pipeline NPV / milestone model. Deep Genomics has no product revenue and no disclosed partnership economics - the value entirely resides in probability-weighted NPVs of clinical programs and partnership milestones. A standard biotech rNPV (risk-adjusted NPV) model is the correct archetype, supplemented by an operating cost (burn) forecast. This is not a SaaS, DTC, or operating-revenue model. - **Forecast horizon & granularity:** - Operating budget: quarterly, 2021–2024 (aligns with stated portfolio milestones and Series C runway). - Pipeline NPV: annual, 2021–2035+ (drug development timelines require 10–15 year horizons to capture peak sales). - **Key drivers & assumptions:** *Pipeline / clinical* - Number of programs entering each stage per year: 2021 actuals and 2022–2024 projections from slide 13 bar chart. - Stage-gate probabilities: preclinical → Phase 1, Phase 1 → Phase 2, Phase 2 → Phase 3, Phase 3 → approval. - Time in each stage:. - Peak sales per program: CNS programs $600M–$1.2B, Metabolic $700M–$1.2B; use midpoints for base case. BioMarin programs: no est. given. - Ramp to peak:. - Royalty rate on partnered programs:. - Profit share / ownership on internal programs:. *Operating costs (burn)* - R&D headcount:. - Platform capex (robotics, compute):. - CRO spend for 28 preclinical programs:. - Clinical trial costs:. - G&A / business development:. *Funding* - Series C: $180M. - Implied runway:. - Non-dilutive partnership inflows:. - **Scenarios (Base / Bull / Bear - which variables flex):** - Base: Stage-gate probabilities at midpoint; 3 of 4 clinic programs reach Phase 2; BioMarin milestone payments on schedule; burn in line with $180M over 3 years. - Bull: DG's 50% preclinical PoS validates in first 2 clinical readouts; additional partnership signed (1–2 programs out-licensed for $50M+ upfront each); complex disease program (Frontotemporal Dementia) achieves Phase 2. - Bear: 2 of 4 clinical programs fail Phase 1; BioMarin partnership narrows; $180M insufficient - bridge raise required in 2023; peak sales estimates revised down 30–40%. - **Required sheets / outputs:** 1. Assumptions dashboard (all drivers in one place, togglable by scenario). 2. Pipeline tracker: program × stage × year × PoS × NPV. 3. rNPV roll-up: probability-weighted DCF per program, summed to portfolio value; discount rate. 4. Operating cost / cash burn model: R&D + G&A + capex, quarterly 2021–2024. 5. Cash bridge: Series C proceeds + partnership inflows − burn = runway. 6. Sensitivity table: portfolio NPV vs. discount rate (10–20%) and clinical PoS (+/−50% vs. base). 7. Partnership economics model: milestone schedule + royalty ramp for BioMarin programs. ## Frequently asked questions ### Is the Deep Genomics financial model free? Yes. The Deep Genomics model is a free Excel download with live formulas. ### Can I change the assumptions? Yes. The workbook is editable and its live formulas recalculate when assumptions change.