AI Computational Biology Math Paper Template
Use this template when a paper connects AI methods, computational biology objects, and mathematical objectives. Keep missing metadata as to verify; do not invent authors, metrics, artifacts, or benchmark details.
Frontmatter
---
title : Paper Title
status : reading
paper :
title : to verify
authors : to verify
venue : to verify
year : to verify
url : to verify
tags :
- papers
---
Summary
One-line question: to verify
Main contribution: to verify
Main axis: AI method / computational biology object / Math objective / benchmark / system / agent workflow
Secondary axes: to verify
Public source: to verify
Reading status: Reading status
Routing
Claim
State claims as paper claims, not facts.
Claim Evidence Status to verifyfigure/table/section to verify to verify
Claim-Type Gate
Select all that apply before summarizing the result.
Claim Type Applies? Required Check architecture improvement to verify fair baseline, ablation, parameter/compute boundary SSL or pretraining to verify pretraining unit, positive/corruption rule, transfer split generative model to verify sampler budget, validity/diversity/novelty/utility, invalid denominator energy or score model to verify energy/score/force definition, sampler or negative process, validity after filtering constrained generation to verify hard constraint vs penalty/projection/repair/filter, invalid-output denominator property or activity prediction to verify label semantics, target/assay context, split unit protein modeling to verify sequence/structure source, PLM/MSA/template policy, family split docking or pose to verify receptor/ligand preparation, pose metric, failed docking handling benchmark result to verify allowed information, selection rule, baseline, uncertainty formula or estimator to verify variables, distributions, objective, estimator assumptions agent or tool workflow to verify tool boundary, verifier, memory, success/failure definition
Evidence level:
evidence_level : mentioned # mentioned | specified | supported | bounded | reproducible
Method
Input object: to verify
Output object: to verify
Representation: to verify
Architecture: to verify
Learning signal: to verify
Objective: to verify
Inference or sampling procedure: to verify
Constraint or filtering rule: to verify
Representation Contract
Field Value Raw unit molecule / protein / residue / atom / complex / text / image / graph / assay row / to verify Model input token / graph / coordinate / embedding / pair feature / latent / to verify Tokenization or graph construction to verifyPositional or geometric encoding to verifyContext source target / pocket / assay / retrieval / template / prompt / not applicable Information available at inference to verifyInformation that must be excluded labels, future data, ground-truth structures, private context, or to verify
Rewrite the key equation with symbol definitions. Use level 3 from Formula explanation ladder when the paper’s claim depends on the equation.
J ( θ ) = E u ∼ q ( u ) [ ℓ θ ( u ) ]
u : to verify
q ( u ) : to verify
ℓ θ ( u ) : to verify
θ : to verify
Explanation level: name / canonical formula / operational form / claim contract / derivation
Relation to evaluation metric: to verify
Objective and Metric Alignment
Field Value Training objective to verifyObjective family likelihood / contrastive / score / energy / velocity / reward / constraint / metric / to verify Model-selection metric to verifyReported metric to verifyClaimed utility to verifyMismatch risk to verify
Computational Biology Contract
Use not applicable only when the paper is not about computational biology.
Field Value Biological object molecule / ligand / protein / pocket / complex / assay / genome region / not applicable Example unit to verifyLabel semantics endpoint, unit, direction, threshold, censoring, source, or not applicable Preprocessing molecule standardization, protein cleaning, coordinate source, or not applicable Split unit scaffold, protein family, complex pair, assay/source, time, template-aware, or to verify Leakage risks duplicate, scaffold, homolog, template, assay/source, coordinate-frame, or to verify
Public Boundary
Field Value Uses only public sources to verify No private paths, servers, accounts, ports, or credentials yes No collaborator-private details yes No unpublished internal results yes Sensitive details removed or generalized to verify
Benchmark
Field Value Dataset or benchmark to verifyTask to verifySplit to verifyMetric to verifyMetric family classification / regression / ranking / probability / generation / pose / enrichment / to verify Baseline to verifyModel-selection rule to verifyUncertainty or seed variation to verifyAllowed information training data, templates, retrieval corpus, prompt examples, or to verify
Evidence and Limits
Concept Updates
Object/entity note to update: to verify
Modality/task note to update: to verify
Architecture or learning note to update: to verify
Math note to update: to verify
Constraint/objective note to update: Constrained optimization , Energy-based model , or not applicable
Evaluation or benchmark note to update: to verify
Potential Korean post route: use the post workflow from the Posts section when a reader-facing essay is needed.
Promotion readiness: AI computational biology math readiness gate