From patient-specific RNA variation to testable ASO candidates. A research platform from Murai Labs that helps teams prioritize antisense-oligonucleotide candidates for splice-altering variants and rare-disease applications.
Given a target gene, exon, and tissue — and optionally a patient's VCF — which ASO candidate should a lab synthesize and test first, and why?
That is the question PINNAL is built to answer. You bring the target; PINNAL returns a ranked, audit-traceable shortlist of antisense-oligonucleotide candidates — each with its mechanism annotation, off-target risk flags, structural accessibility, and a machine-readable lab protocol. It is a candidate ranker, not a target-discovery tool, and each ranking comes with the reasons behind it.
Nucleic-acid therapeutics target the transcript, not the protein — the fastest-growing path to drugging the previously undruggable.
Antisense oligonucleotides and splice-switching oligos already work in the clinic. Nusinersen rescues SMN2 exon-7 inclusion in spinal muscular atrophy; milasen was designed and dosed for a single child. The science is proven. The bottleneck is selection: for any target gene, exon, and tissue there are 200–5,000 possible binding sites, each synthesis costs $5K–$50K, and the field still chooses which ~10 to make first by intuition, tiling, and a literature scan.
For N-of-1 and rare-variant cases the stakes are sharper — a patient's own VCF can change which candidate is right, and no off-the-shelf tool weighs all of it at once. PINNAL is built for exactly that decision.
ASO design still has a translation gap.
Patient-specific ASO development means stitching together many steps — variant interpretation, splice-mechanism review, sequence analysis, target-region selection, off-target screening, structure checks, manual candidate triage, and protocol prep — often across a different tool for each. The decision trail fragments, and candidate selection comes down to spreadsheets, expert judgment, and manual handoffs.
For individualized medicine that fragmentation matters. Every candidate tested consumes time, material, assay capacity, and patient-specific opportunity.
PINNAL helps prioritize what to test next.
It combines target context, sequence-level features, splice-relevant signals, off-target risk flags, accessibility checks, and experimental feedback into a single ranked candidate shortlist — built for researchers and translational teams moving from a molecular hypothesis to a practical experimental plan.
- Identify candidate ASO binding regions around a target exon or splice-relevant locus.
- Rank candidates using multiple evidence signals rather than a single score.
- Review why each candidate was prioritized.
- Compare tradeoffs between predicted activity, specificity, accessibility, and confidence.
- Generate protocol-ready outputs for early wet-lab testing.
- Ingest experimental results to inform the next round of prioritization.
- Support N-of-1 workflows starting from a patient-specific variant file.
How can PINNAL choose ASO candidates?
- Repeat for each candidate site
- Also consider accessibility and off-target risk
Not just a score — a decision trail.
ASO prioritization is rarely a single-variable problem; a candidate can look promising for one reason and risky for another. PINNAL makes those tradeoffs visible, comparing candidates across multiple dimensions and preserving an audit trail from input target to ranked output — easier to review, reproduce, and discuss across computational, bench, clinical, and translational teams.
Built for patient-specific ASO research.
For N-of-1 and rare-disease work, the key question is often whether a patient-specific variant creates a splice defect that an ASO can rescue. PINNAL lets teams include patient variant information during prioritization, then compare a general target run with a patient-specific run to surface candidates more relevant to that patient's molecular context.
The first experiment should inform the next one.
ASO development is iterative; the first screen rarely ends the story. PINNAL captures experimental results and uses them to support the next round of prioritization, moving teams from one-off candidate lists toward a learning workflow that improves with real experimental outcomes.
| Target | Context | Why it matters | Early result |
|---|---|---|---|
| ISS-N1 / SMN2 | Spinal muscular atrophy | Benchmark for exon-inclusion rescue through silencer blockade | 3 of 3 target-overlapping windows in PINNAL's top 10, in a focused search of ~41 overlapping ASO binding windows |
| AR-V7 / CE3 | Prostate cancer | Benchmark for cryptic-exon exclusion in disease-associated splicing | Top-10 enrichment 1.64, against 0.82 for Enformer or Borzoi alone |
| MFSD8 / Milasen | Rare disease / N-of-1 | Benchmark for patient-specific splice rescue | Top-10 enrichment 1.64; Enformer alone performed better at 2.46 |
| AP4E1 | Enhancer-driven cryptic exon | Tests whether experimentally effective ASOs rank first | Found the relevant enhancer region; the published ASO ranked 111th |
| ABCA4 | Published ASO series | Tests whether effective ASOs separate from less effective ones | Covered the region containing effective ASOs; ordering was weak (AUC 0.58) |
Small, focused tests. These scores measure target overlap, not biological activity; candidate activity was not tested in cells. The next challenge is improving which candidates reach the top.
Clear scope matters.
PINNAL is not a clinical diagnostic platform. It does not replace clinical genetic interpretation, ACMG review, RNA-seq confirmation, minigene assays, patient-cell validation, toxicology, GMP manufacturing, regulatory review, or clinical decision-making.
It is intended to support research-stage ASO prioritization and translational planning. Final candidate selection and clinical use require expert review and experimental validation.
Preethi Ravindranathan
Preethi Ravindranathan brings experience across molecular biology, RNA biology, oncology drug discovery, biomarker research, and clinical research operations.
Her research background includes gastrointestinal cancer biomarker discovery, RNA-seq and non-coding RNA biology, androgen receptor signaling, peptidomimetic drug discovery, and RNA-targeted therapeutic strategies, including ASO and siRNA approaches to cancer-associated splice variants.
At UT Southwestern Medical Center, she was first author on a Nature Communications publication on peptidomimetic targeting of androgen receptor–coregulator interactions in prostate cancer, with related co-authored work in PNAS and Oncotarget.
At Murai Labs, Preethi leads PINNAL's clinical and translational strategy, helping connect computational ASO candidate prioritization with rare-disease workflows, experimental validation, and patient-specific research applications.
- Pre-alpha, demo-ready platform.
- N-of-1 patient-VCF intake in the workflow.
- Lab partnerships in outreach — wet-lab readouts calibrate the ranking, the partnership unlock.
We are looking for early scientific collaborators.
PINNAL is being developed for researchers and translational teams working on splice-modifying ASOs, rare-disease mechanisms, and individualized medicine. We are especially interested in collaborations where computational prioritization can be paired with real experimental feedback.
Ideal collaboration areas
- Patient-specific splice-altering variants
- Pseudoexon or cryptic-exon rescue
- Exon inclusion or exon skipping programs
- Minigene or patient-cell assay systems
- Retrospective ASO screen datasets
- Transcriptome-wide off-target profiling
- Rare-disease N-of-1 workflows
What collaborators can expect
- A structured candidate-prioritization workflow
- Ranked ASO candidate shortlists
- Evidence-linked candidate review
- Protocol-ready outputs for early screening
- Support for feedback-driven iteration
- Transparent discussion of limitations and assumptions
Have a tested splice-modifying ASO series? Let's compare. We are looking for ASO sequences with measured outcomes, successful and unsuccessful candidates, and the chemistry, dose, cell model and assay details. Write to [email protected].
All outputs are intended for research use and require expert review and experimental validation. You can also contact Murai Labs to discuss a target, dataset, or collaboration.
KALAVAI - Cooperative Specialist Fusion via Frozen-Layer MoE Routing.