Operating plan · rev 2 · Aug 2026 Target 80% · 90% does not exist

The India
Route

Building autism-therapeutics capability from Bengaluru instead of buying it in New York — through open courses, Indian government grants, national-lab tie-ups and contract research.

Capability coverage against the 14 weighted disciplines

66%Self-study ceiling

80%This plan

85%Mt Sinai PhD

90%Nobody

Courses & open data — free, yours to keep

Grants, labs & CROs — contracted, not owned

Asymmetry 01

CNS chemistry, 8.5 vs 5.0

Syngene, Aragen, Sai Life and Jubilant do CNS medicinal chemistry and DMPK for global pharma at a standard no US academic centre matches. On the discipline Seaver is weakest at, you would be stronger.

Asymmetry 02

An unbuilt cohort

India's largest published autism trio study is ~137 trios. MSSNG holds 13,800+ genomes. Polygenic scores lose accuracy in South Asians. The cohort that fixes this does not exist and is buildable here.

Months 0–3

Foundation

Nothing here adds a coverage point. It converts the 66% you can reach alone from theoretical into executable, and settles the legal questions before they block a grant.

₹5.1 L
6 h / week
+0 pts

Build the dry lab. Ryzen 9 or Threadripper, 128–256 GB RAM, RTX 5090-class GPU, 40 TB NAS. ~₹5 L capex, or ₹20–50 k/month on cloud GPU instead. This is the highest-return spend on the whole page.
Install and verify the free stack end-to-end: GATK, RDKit, PyMOL, Galaxy, Bioconductor.
Confirm grant eligibility before anything else. BIRAC BIG caps company age at five years — check that against your incorporation date now, because it may be a closing window rather than an open one.
Confirm whether your incubator is one of BIRAC's recognised BIG Partners. If not, identify which one you would route through.
Apply for DSIR SIRO recognition — exempts customs duty on imported research equipment and is a prerequisite for several schemes.
Register for the first two courses in Phase 01 and for NIH IPPCR.

Gate — do not proceed until

You can run a variant-calling pipeline end-to-end on public data without following a tutorial, and you know in writing whether your entity is BIG-eligible.

Months 1–15 · runs under everything else

Knowledge

Two courses in parallel, continuously. The single highest-value item is not on NPTEL — it is free, from the NIH, and fixes the discipline weighted second-highest in the whole framework.

₹45,000
10 h / week
0 → 66 pts

Do these first — outside NPTEL
Course	Source	Load
Introduction to the Principles & Practice of Clinical Research~40 lectures. Study design, biostatistics, protocol prep, monitoring, ethics, regulatory. Certificate at 75%. Free. Moves trials & measurement from 3 → 6.	NIH OCRECO	Self-paced
TeachOpenCADDOpen Jupyter notebooks. Pushes cheminformatics to 9.	Volkamer Lab	Self-paced
Galaxy Training Network	Galaxy	Self-paced
EMBL-EBI Train OnlineEnsembl, UniProt, ChEMBL, AlphaFold.	EMBL-EBI	Self-paced
MPG Primer lecture seriesFree statistical-genetics lectures from the people who built gnomAD and Hail — the exact toolset the Phase 02 paper runs on.	Broad Institute	Self-paced
Practical Cheminformatics Tutorials24 Colab notebooks by Pat Walters. Pairs with TeachOpenCADD; the industry-standard idiom for RDKit and ML on molecules.	Pat Walters	Self-paced
Neuromatch AcademyComputational neuroscience intensive. Runs each July; applications close in March — plan for the 2027 cohort.	Neuromatch	3 weeks

NPTEL stack — enrol via Swayam, ₹1,000 per certificate
Course	Institute	Weeks
Tier A — core
Computational Systems Biology	IIT Madras	12
BioInformatics: Algorithms & Applications	IIT Madras	12
Next Generation Sequencing Technologies	IIT Kharagpur	12
Computational Genomics	IISER Bhopal	12
Tier B — the chemistry inversion, where NPTEL beats the PhDs
Computer Aided Drug DesignDocking, QSAR, pharmacophore, ADME — taught entirely with free software.	IIT Madras	8
AI in Drug Discovery and DevelopmentPredictive modelling, generative design, repurposing.	IIT (BHU)	12
Medicinal Chemistry	IISER Pune	12
Tier C — neuro and structure
Computational Neuroscience	IIT Kharagpur	12
Algorithms for Protein Modelling & Engineering	IIT Kharagpur	12
Statistics for Biomedical Engineers	IIT Madras	12
Optional depth
Functional Genomics	IIT Kanpur	4
Introduction to Proteogenomics	IIT Bombay	12
Algorithms in Computational Biology	IISc Bengaluru	12

Checked against the field — budget ₹25 k per course, certificates now in scope
Global option	Versus	Verdict
Systems Biology & Biotechnology SpecializationIcahn School of Medicine at Mount Sinai — Ma'ayan, Iyengar, Sobie. The benchmark institution's own curriculum: network modelling, omics, single-cell. Certificate included in Coursera Plus.	Computational Systems Biology	Supplement
HarvardX Data Analysis for Life SciencesRafael Irizarry. Statistics and linear models in R, built for genomics data. Verified certificates ~$200 per course — buy for the two statistics courses, audit the rest.	Statistics for Biomedical Engineers	Swap
MIT 6.047/6.878 Computational BiologyManolis Kellis. Full lectures free on OCW and YouTube; deeper algorithmic treatment.	BioInformatics: Algorithms	Supplement
UCSD Bioinformatics SpecializationPevzner & Compeau, problem-driven via Rosalind. Certificate included in Coursera Plus.	Computational Genomics	Supplement
Wellcome Connecting ScienceFree short courses on genomic analysis and interpretation.	NGS Technologies	Supplement
Computational NeuroscienceUniversity of Washington — Rao & Fairhall. The classic MOOC treatment; Python/Matlab exercises. Certificate included in Coursera Plus; Neuromatch remains the deeper option.	Computational Neuroscience	Supplement
DavidsonX Medicinal ChemistryBuilt with Novartis. 7 weeks on receptors, pharmacokinetics, metabolism, structure–activity. Do it as a primer before the IISER Pune course; verified certificate ~$99 if wanted.	Medicinal Chemistry	Supplement
Schrödinger: Intro to Molecular Modeling in Drug Discovery$140–150 at student rate — inside budget. Industry-recognised certificate with hands-on web Maestro. The advanced tiers ($825) stay out of scope.	CADD	Add
CADD & AI-in-drug-discoveryChecked against Schrödinger certification (paid, proprietary stack), MolSSI, CCDC and the workshop market: nothing at any price under ₹25 k teaches docking, QSAR and generative design on open tools with a proctored certificate. NPTEL stays the core; the Schrödinger intro adds the industry-tool layer on top.	CADD · AI-DD	NPTEL wins

Gate — do not proceed until

IPPCR certificate in hand, six NPTEL certificates, and you can read a published Phase 2 autism trial protocol and name three things you would change about its endpoints.

Months 6–18 · overlaps Phase 01

Credibility

Certificates open no doors on their own. One preprint using open data does. Produce a citable artifact before asking any institution, funder or partner for anything.

₹2–5 L
12 h / week
Converts 66 pts

Apply for access to MSSNG — 13,800+ autism whole genomes, free to researchers, with local compute and storage.
Register for SFARI Base (SPARK, Simons Simplex Collection) and pull reference frequencies from gnomAD.
Obtain IndiGen data from CSIR-IGIB — 1,029 Indian whole genomes, the only South Asian baseline that exists.
Write one paper. Suggested question: apply an SFARI/ClinGen ASD gene panel across MSSNG and IndiGen and quantify how South Asian allele frequencies shift variant interpretation. Real gap, computationally tractable, no wet lab, no permission needed. This is also the exact evidence a supervisory letter or an MoU request needs behind it.
Post to medRxiv, then submit for peer review.
Recruit one Indian academic co-author before submission — NIMHANS, IGIB or CCMB. The co-author matters more than the journal.
Submit an abstract to INSAR. It is where every name in this field is in one room.

Gate — do not proceed until

One preprint posted and one Indian academic co-author has signed on. If no academic will co-author after eighteen months, the binding constraint is network, not knowledge — go solve that instead of continuing.

Months 12–24

Money

BIRAC first, because it is non-dilutive, sized right for proof of concept, and the only one that takes a company rather than a professor. Everything larger routes through an academic partner.

₹50 L in
8 h / week
in windows

BIRAC Biotechnology Ignition Grant — ₹50 L grant-in-aid over 18 months, non-equity, disbursed against 3–4 milestones. Read the scheme guidelines and watch open calls; windows run ~45 days. Eligibility: Indian citizen, either incubated or a registered company with a functional R&D lab, company age under five years.
Cross-check the scheme summary on the ISTI portal — it states eligibility in plainer terms than the guidelines PDF.
Startup India Seed Fund Scheme — ₹20 L grant for proof of concept, plus up to ₹50 L as convertible debt for market entry. Runs in parallel with BIRAC, different selection pool.
Scan DBT calls continuously. Most large DBT and ANRF instruments require an institutional principal investigator, so your Phase 02 co-author becomes the PI and you become the industry partner.
Once BIG delivers a proof of concept, step up to BIRAC SBIRI or BIPP — ₹1–5 crore, industry-led, designed for exactly this transition.

Gate — do not proceed until

BIG awarded, or two refusals with written feedback you have actually addressed. Two refusals means the proposition is not fundable as framed — rewrite it or route it through the partner institution, do not simply resubmit.

Months 18–36

Partnerships

Rent, do not build. A Tier-2 iPSC suite is ₹1–3 crore of capex plus a regulatory problem you cannot solve as an individual. An MoU with a national lab costs nothing and the partner holds the stem-cell committee.

₹50 L–3 Cr
15 h / week
66 → 80 pts

National labs — MoU route, zero capex
Institution	What it gives you
BRIC-inStem, Bengaluru	iPSC and organoids. Critically, they hold the registered stem-cell committee, which you cannot constitute as an individual.
NIMHANS, Bengaluru	The one that matters most. Autism clinical population, neuroimaging, child psychiatry, and the route to a cohort.
CSIR-IGIB, Delhi	IndiGen, Indian population genomics, CRISPR diagnostics.
CSIR-CCMB, Hyderabad	Genomics at scale.
NCBS, Bengaluru	Neural development, imaging, cryo-EM access.
NBRC, Manesar	Systems and circuit neuroscience.

Contract research — where India outperforms Mount Sinai
Company	Scope
Syngene	Discovery through development, integrated.
Aragen Life Sciences	CNS medicinal chemistry, DMPK.
Sai Life Sciences	Med chem, lead optimisation.
Jubilant Biosys	Discovery chemistry and biology.
Veeda Clinical Research	Clinical phase execution.

Commercial whole genome sequencing runs ₹25–60 k per genome in India — MedGenome, Neuberg, Strand, Eurofins. No reason to own a sequencer.
Register the regulatory picture early: CDSCO for trial approval, CTRI for registration, and the National Guidelines for Stem Cell Research. Governance shifted in March 2024 — NAC-SCRT was dissolved and stem-cell proposals now route through Institutional Ethics Committees with at least two stem-cell experts, one external to the institution. Verify the current position with ICMR and DBT before planning around it.
The precedent: Mount Sinai's Seaver Center supplies iPSCs and contracts the organoid phenotyping out. The best-funded autism centre on earth does not own the whole stack either.

Gate — do not proceed until

Two signed MoUs and one CRO scope of work executed and delivered. A memorandum nobody has run an experiment under is not a partnership.

Year 3 onward

The cohort

The point of the previous five years. Not a coverage point — the differentiated asset that makes international collaboration come to you rather than the other way round.

₹20–50 Cr
Full time
Institution-scale

Target 1,000+ deeply phenotyped autism trios through NIMHANS and partner sites. Existing Indian screening work has run multi-stage pipelines across tens of thousands of children, so ascertainment capacity exists.
Whole genome, not exome. The noncoding layer is where variant-effect prediction has the most to gain, and it is what MSSNG did right. 3,000 samples at ₹25–60 k = ₹7.5–18 Cr.
Layer AI-based digital phenotyping over the clinical assessments — video behavioural coding, eye tracking, vocal prosody. This is the highest-scored AI application in the whole framework, it needs cohorts and ML rather than a wet lab, and you would have both.
Co-fund internationally. Autism Speaks, SFARI and the DBT/ICMR national programmes all have reasons to want South Asian representation. Do not self-fund this.
Publish the cohort as an open resource. MSSNG's openness is why it became the field standard.

Gate — the real one

If the cohort cannot attract international co-funding, it is not wanted badly enough to justify ₹20 crore of your own risk. That answer is worth finding out in year three rather than year seven.

₹

Cumulative

Funds

Personal outlay stops at roughly ₹10 lakh. Everything past that is grant-funded or does not happen.

Phase	Item	Out	In
00	Workstation, storage	₹5.0 L	—
01	NPTEL ×10, Coursera Plus, Schrödinger, edX certs	₹0.5 L	—
02	Cloud, conference, submission	₹3.0 L	—
03	BIRAC BIG	—	₹50 L
03	SBIRI / BIPP	—	₹1–5 Cr
04	CRO chemistry campaign	₹2–10 Cr	—
05	Cohort sequencing, phenotyping	₹7.5–18 Cr	—
05	International co-funding	—	Required
—	Personal risk capital, phases 00–02	₹8.5 L	—

⏱

Sustained load

Time

Ten hours a week for two years is the whole knowledge phase. That is one weekday evening and one weekend morning. It is small enough to be real and large enough to matter — which is exactly why it fails quietly if you do not defend it.

Period	Hours / week	Occupied by
Months 0–3	6	Setup, two courses running
Months 3–15	10	Two courses in parallel, continuously
Months 6–18	+12	Analysis and writing, overlapping courses
Months 12–24	+8	Grant windows only, ~45 days each
Months 18–36	15	Partnership management, CRO oversight
Year 3+	Full time	This stops being a side pursuit

Stop conditions

When to abandon this

A plan without exits is a wish. Each of these is a real signal, not a discouragement — finding out early is the entire value.

Month 12, fewer than four NPTEL courses done. The ten-hour budget is fictional. Cut scope to the AI and measurement lane only — it is the highest-weighted thing you can do without partners, and it is the one you are already good at.
Month 18, no Indian academic co-author. The constraint is network, not knowledge, and no additional certificate fixes it. Solve access directly or stop.
Two BIG refusals with feedback addressed. The proposition is not fundable as framed by people whose job is funding exactly this. Rewrite the thesis or route it through the partner institution as the applicant.
No international interest in the cohort by year four. Do not self-fund a ₹20 crore resource that the field has declined to co-fund.
At any point, if the enterprise reaches 80% while you personally stay near 66%. That gap means you are buying results you cannot evaluate. The fix is a scientific co-founder, not more contracts.