name: mutation-kinetics-miner
description: This skill should be used when the user asks to "extract mutation-kinetics from papers", "analyze protein mutations and substrate relationships", "enzyme mutation literature mining", "논문에서 변이-동역학 관계 추출", "단백질 변이와 기질 관계 분석", or needs systematic extraction of mutation-property-mechanism relationships from scientific literature for protein engineering decisions.
user_invocable: true

Mutation-Kinetics Miner

A comprehensive literature mining skill for extracting mutation-property-mechanism relationships from protein engineering publications. This skill systematically collects and structures data on how specific amino acid changes affect enzyme kinetics, stability, and specificity, along with mechanistic explanations.

When to Use This Skill

Use this skill when:
- Investigating how specific mutations affect enzyme catalytic properties
- Designing protein engineering experiments based on literature evidence
- Systematically reviewing mutation effects for a target enzyme
- Understanding mechanistic explanations for observed property changes
- Building a knowledge base of mutation-function relationships

Input

Provide the target protein using one of these formats:

/mutation-kinetics-miner Lipase B
/mutation-kinetics-miner P06278
/mutation-kinetics-miner EC 3.1.1.3

Accepted input formats:
- Protein name: "Lipase B", "Cytochrome P450", "TEM-1 beta-lactamase"
- UniProt ID: "P06278", "Q9Y6K9"
- EC number: "EC 3.1.1.3", "3.4.21.4"

Note: For ambiguous protein names, provide UniProt ID for precise identification.

Core Workflow (4 Phases)

Phase 1: Target Protein Identification

Establish comprehensive protein context before literature search.

Step 1.1: Retrieve Protein Information

Invoke scientific-skills:uniprot-database to retrieve:
- Canonical sequence and length
- Active site residues and binding sites
- Known natural variants and disease mutations
- Protein family classification

Step 1.2: Obtain Baseline Kinetics

Invoke scientific-skills:brenda-database to collect:
- Wild-type Km, kcat, kcat/Km values
- Substrate specificity profile
- Optimal pH and temperature
- Known inhibitors and activators

Step 1.3: Structural Context

Invoke scientific-skills:alphafold-database or scientific-skills:pdb-database:
- Retrieve available structure (experimental or predicted)
- Identify active site geometry
- Map catalytic residues in 3D space

Output: Protein Profile

## Protein Profile: [Name]

| Property | Value | Source |
|----------|-------|--------|
| UniProt ID | P12345 | UniProt |
| EC Number | 3.1.1.3 | BRENDA |
| Length | 534 aa | UniProt |
| Active Site | Ser105, His224, Asp187 | UniProt |
| Wild-type Km | 0.45 mM (p-NPB) | BRENDA |
| Wild-type kcat | 120 s⁻¹ | BRENDA |
| Optimal pH | 7.5 | BRENDA |
| Structure | 1TCA (2.3 Å) | PDB |

Phase 2: Literature Search & Collection

Systematically search multiple sources for mutation studies.

Step 2.1: PubMed Primary Search

Invoke scientific-skills:pubmed-database with structured query:

"[protein name]" AND (mutation OR mutant OR "site-directed mutagenesis" OR
"directed evolution" OR "rational design") AND (kinetic* OR Km OR kcat OR
"catalytic efficiency" OR "substrate specificity" OR thermostabil*)

Retrieve:
- PMID, DOI
- Title, authors, journal, year
- Abstract text

Step 2.2: Preprint Search

Invoke scientific-skills:biorxiv-database for recent unpublished work:
- Same query structure as PubMed
- Flag as preprint in output

Step 2.3: Citation Expansion

Invoke scientific-skills:openalex-database to:
- Find citing papers (forward references)
- Find cited papers (backward references)
- Identify review articles summarizing mutation data

Stopping Criteria (to prevent scope explosion):
- Maximum 2 citation generations (papers citing papers)
- Maximum 50 additional papers from citation expansion
- Stop when <10% of new papers contain relevant mutations
- Prioritize: reviews > high-citation papers > recent papers

Step 2.4: Deduplication

Before proceeding to Phase 3, deduplicate the paper collection:
1. Remove exact duplicates by DOI/PMID
2. Merge entries from different databases (same paper, different metadata)
3. Flag potential duplicates (same authors, same year, similar titles)
4. Record final count: unique papers / total retrieved

Filtering Criteria:

Output: Curated Paper List

## Literature Collection: [N] papers

| PMID | Year | Title | Mutations | Data Types |
|------|------|-------|-----------|------------|
| 12345678 | 2023 | Engineering thermostable... | 15 | Km, kcat, Tm |
| 23456789 | 2022 | Substrate specificity of... | 8 | Km, specificity |

Phase 3: Relationship & Mechanism Extraction

Extract structured data from each paper.

Extraction Method:
- Use scripts/extract_mutations.py for automated mutation notation parsing from text
- Manual curation required for mechanism interpretations and context validation
- Use Phase 1 structure (cached) for position mapping—do not re-retrieve

Limited Access Protocol (for paywalled papers):

When full-text is unavailable:
1. Extract mutation notations and fold-changes from abstract
2. Flag as "abstract-only" in output with reduced confidence
3. Add to priority list for full-text retrieval
4. Do not extract mechanism details (require full-text Methods/Results)

Step 3.1: Mutation Notation Extraction

Identify all mutation notations:
- Standard format: D121N, A234S, W167F
- Multiple mutations: D121N/A234S (double mutant)
- Insertions/deletions: ΔLoop, +Gly insert

Position Numbering Caution:

Position numbers often differ between sources. Always verify:

Verification protocol:
1. Align paper sequence to UniProt canonical
2. Confirm position matches expected residue type
3. If mismatch, recalculate using sequence alignment
4. Flag uncertain positions with "[position unverified]"

Step 3.2: Substrate Identification

Extract substrate information:
- Substrate name and structure (if SMILES provided)
- Natural vs synthetic substrates
- Substrate analogues used in assays

Step 3.3: Kinetic Parameter Extraction

Extract quantitative values WITH assay conditions (critical for comparability):

Assay Conditions (Mandatory):

Note: Values measured under different conditions cannot be directly compared. Flag conflicting values with condition context.

Step 3.4: Mechanism Explanation Extraction

Capture mechanistic insights:

Step 3.5: Additional Property Extraction

Phase 4: Synthesis & Output

Generate structured output formats. Include extraction metadata for reproducibility.

Extraction Metadata (Include in all outputs):

---
Extraction Date: YYYY-MM-DD
Skill Version: mutation-kinetics-miner v1.1
Target Protein: [Name] (UniProt: [ID])
Papers Analyzed: [N] unique / [M] total retrieved
Query: "[search terms used]"
---

Output Format 1: Kinetics Change Table (Required)

## Kinetic Parameter Changes

| Mutation | Position | Substrate | Km (mM) | kcat (s⁻¹) | kcat/Km | Fold Change | Conf. | PMID |
|----------|----------|-----------|---------|------------|---------|-------------|-------|------|
| D121N | Active site | Glucose | 2.3 | 450 | 1.96×10⁵ | -2.1× | High | 12345678 |
| A234S | Surface | Glucose | 0.42 | 130 | 3.10×10⁵ | +1.5× | Med | 12345678 |
| W167F | Substrate pocket | Maltose | 5.1 | 890 | 1.74×10⁵ | +3.2× | Low | 23456789 |

**Confidence levels:**
- **High**: ≥3 independent sources, consistent values, similar assay conditions
- **Med**: 1-2 peer-reviewed sources, complete assay conditions reported
- **Low**: Single source, preprint, abstract-only, or conflicting values

Output Format 2: Mechanism Analysis (Required)

## D121N Mutation Analysis

### Structural Changes
- Asp121 carboxyl group → Asn121 amide group
- H-bond network weakened (distance: 2.8Å → 3.5Å)
- Active site polarity reduced

### Substrate Binding Effects
- Glucose binding affinity decreased 2.1× (Km: 1.1 → 2.3 mM)
- Transition state stabilization impaired
- Binding entropy penalty increased

### Catalytic Mechanism Changes
- Proton relay pathway altered
- General base catalysis efficiency reduced
- kcat decreased (890 → 450 s⁻¹)

### Other Property Changes
- Thermal stability: Tm unchanged (72°C)
- pH optimum: shifted 7.0 → 7.5
- Solubility: improved 1.5×

### Source
- PMID: 12345678 (Kim et al., 2023)
- DOI: 10.1016/j.jmb.2023.xxx

Output Format 3: Property Summary Table (Optional)

## Property Change Summary

| Mutation | Tm (°C) | pH Opt | Solubility | Specificity | Mechanism Summary |
|----------|---------|--------|------------|-------------|-------------------|
| D121N | 72 (=) | 7.0→7.5 | +1.5× | Maintained | H-bond weakening → Km increase |
| A234S | 68→75 | 7.0 (=) | +2.0× | Broadened | Surface hydrophilicity → stability |
| W167F | 72 (=) | 7.0 (=) | = | Shifted to maltose | Pocket volume → new substrate fit |

Output Format 4: Engineering Recommendations (Optional)

## Engineering Insights

### Beneficial Mutations
| Mutation | Benefit | Trade-off | Combinable |
|----------|---------|-----------|------------|
| A234S | +7°C Tm | Slight Km increase | Yes (surface) |
| T45V | +2× kcat | None observed | Yes |

### Positions to Avoid
| Position | Reason | Evidence |
|----------|--------|----------|
| Asp121 | Essential for catalysis | All mutations reduce kcat >10× |
| His224 | Catalytic triad | Mutations abolish activity |

### Suggested Combinations
Based on structural analysis and literature:
1. A234S + T45V: Stability + activity (no structural conflict)
2. W167F for substrate switching (incompatible with WT specificity)

Output Format 5: Machine-Readable Export (Optional)

For downstream computational analysis, export as CSV/TSV:

mutation,position,position_type,substrate,km_mM,kcat_s-1,kcat_km,fold_change,confidence,ph,temp_C,buffer,pmid,doi,notes
D121N,121,active_site,glucose,2.3,450,1.96e5,-2.1,high,7.5,25,50mM_Tris,12345678,10.1016/xxx,
A234S,234,surface,glucose,0.42,130,3.10e5,1.5,med,7.5,25,50mM_Tris,12345678,10.1016/xxx,
W167F,167,substrate_pocket,maltose,5.1,890,1.74e5,3.2,low,7.0,37,50mM_phosphate,23456789,,abstract_only

CSV Schema:
- All numeric fields use standard units (mM, s⁻¹, M⁻¹s⁻¹)
- Empty fields = not reported
- notes field for flags: abstract_only, preprint, position_unverified

Scientific Skills Integration

When to use ESM prediction:
- Mutation not found in literature → predict effect computationally
- Validate extracted effects → cross-check with ESM scores
- Combinatorial mutants → predict effects of untested combinations
- Novel positions → assess conservation and predicted impact

Validation Checklist

Before finalizing output:

• [ ] Protein profile includes active site and WT kinetics
• [ ] Literature search covers PubMed + preprints
• [ ] All mutation notations follow standard format (X###Y)
• [ ] Kinetic values include units and error bars where available
• [ ] Mechanism explanations cite specific structural features
• [ ] Source citations (PMID/DOI) provided for all data

Troubleshooting

Handling Conflicting Data

When multiple sources report different kinetic values for the same mutation:

1. Record all values with conditions:
   D121N Km: 2.3 mM (pH 7.5, 25°C) [PMID:123] D121N Km: 3.1 mM (pH 8.0, 37°C) [PMID:456]

2. Prioritization order:

3. Peer-reviewed > preprint
4. Larger sample size (n>3) > smaller
5. More recent publication > older

6. Consistent methodology > mixed

7. Report as range when conditions differ:
   D121N Km: 2.3-3.1 mM (depends on pH/temp)

8. Flag high-confidence vs uncertain:

9. High confidence: ≥3 independent reports, similar conditions
10. Uncertain: single report or conflicting values

Additional Resources

Reference Files

• references/search-queries.md - Pre-built PubMed query templates
• references/mutation-patterns.md - Common mutation notation formats

Scripts

• scripts/extract_mutations.py - Parse mutation notations from text