top of page

SCIENTIFIC FOUNDATIONS

Resolve what the evidence actually supports.

Biology-First Intelligence (BFI) is AnnieGuard’s proprietary causal reconstruction methodology for resolving difficult biological questions when heterogeneous evidence remains compatible with multiple explanations. BFI determines what is supportable, what remains unresolved, and what additional evidence would most change the scientific conclusion.

COMPLEX EVIDENCE

Genomic · Molecular · Clinical · Longitudinal · Multimodal

BIOLOGY-FIRST INTELLIGENCE

Supported

Unresolved

Next Evidence

THE R&D PROBLEM

Strong signals can still support the wrong biological conclusion.

Genomic, molecular, clinical, and computational evidence can produce compelling signals while leaving multiple causal interpretations consistent with the data. For R&D teams, that ambiguity affects target selection, biomarker strategy, mechanism assessment, experimental prioritization, and translational decisions.

Prediction can rank possibilities. BFI is designed to determine what the evidence can actually support.

01

RESOLVE

What is supported?

Evaluate competing biological interpretations against the available evidence.

02

PRESERVE

What remains unresolved?

Keep uncertainty and competing explanations explicit when unique identification is not supported.

03

DISCRIMINATE

What evidence matters next?

Identify the evidence gap most relevant to resolving the remaining ambiguity.

Uncertainty is a scientific output.

SCIENTIFIC DIFFERENTIATION

A different question produces a different scientific output.

PREDICTIVE / ASSOCIATIONAL

What predicts or is associated with an outcome?

Output → prediction, association, classification, risk.

CAUSAL INFERENCE

What causal relationship or effect is identifiable under specified assumptions?

Output → causal estimates or intervention-related conclusions.

SYSTEMS BIOLOGY

What pathways, networks, or system-level relationships characterize the biology?

Output → network organization and mechanistic hypotheses.

BIOLOGY-FIRST INTELLIGENCE

What biological architecture can the available evidence actually support?

Output → supportable biological state, unresolved alternatives, mechanistic interpretation, and discriminating evidence.

A plausible explanation is not automatically an identified explanation.

EVIDENCE IN APPLICATION

21-case undifferentiated sarcoma study

BFI was applied across 21 independently frozen cases to test whether heterogeneous molecular evidence converges at a higher biological level—and whether reconstruction generates information beyond exact feature matching.

19 / 19

Molecularly informative cases supported multi-event structural-genome remodeling

18 / 18

Claim-informative cases supported a broad viable dosage-imbalanced / aneuploid state

18 / 19

Generated at least one state-level mechanistic hypothesis beyond exact feature matching

These results are supported by the public case-study artifact. 

The cohort did not converge on a universal driver. The first stable convergence emerged at the biological process/state level.

For R&D teams, that changes the question from “Which alteration should we prioritize?” to “Which biological organization does the evidence support?”

VALIDATION

Algorithmic Integrity & Structural Consistency

BFI has completed internal methodological validation across its tested registered execution scope. Validation was designed to assess deterministic execution, root-identification recovery, model/equivalence preservation, and monotonic behavior relative to evidence strength.

STAGE 01

Internal Methodological Validation

Status: Complete

Algorithmic & causal integrity established within the tested registered execution scope.

STAGE 02

Cohort Application Evidence

Status: Demonstrated

Applied across a 21-case undifferentiated sarcoma cohort, demonstrating recurrent higher-order biological organization and reconstruction-derived mechanistic information beyond exact feature matching.

STAGE 03

External Biological / Translational Validation

Status: Partnering stage

Positioned for external biological validation and wet-lab hypothesis testing through structured pharmaceutical R&D collaborations.

PHARMACEUTICAL R&D

Where BFI can inform decisions

Target & Mechanism Evaluation

Determine whether a target or proposed mechanism reflects broader biological organization or a local molecular feature.

Biomarker Strategy

Assess whether a biomarker corresponds to a coherent biological state.

Patient Stratification

Evaluate whether molecularly heterogeneous cases converge on shared disease organization.

Experimental Prioritization

Identify evidence most capable of resolving a consequential biological uncertainty.

Translational Decision Support

Clarify what the current evidence supports before committing additional R&D resources.

The value is not another target list.

It is a better-defined biological decision boundary.

Bring AnnieGuard a difficult target, biomarker, mechanism, cohort, or multimodal evidence question. We’ll determine whether the available evidence supports the conclusion—and where the remaining uncertainty lies.

Defined research engagements · MSA / SOW compatible

bottom of page