スキル一覧に戻る

interpret-chromatogram

pjt222
更新日 Yesterday
2 閲覧
17
2
17
GitHubで表示
デザインdesign

について

このClaudeスキルは、GC/HPLCクロマトグラムを解析し、システム適性を検証し、保持時間とスペクトルマッチングによるピーク同定、品質評価を伴う積分処理を行います。定量分析の信頼性を確保するための性能指標を計算し、クロマトグラフィーデータを処理する開発者を支援します。データのレビュー、分析方法の適合性確認、ピーク積分の問題解決時にご利用ください。

クイックインストール

Claude Code

推奨
メイン
npx skills add pjt222/agent-almanac -a claude-code
プラグインコマンド代替
/plugin add https://github.com/pjt222/agent-almanac
Git クローン代替
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/interpret-chromatogram

このコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします

ドキュメント

Interpret a Chromatogram

Systematic interpret GC/HPLC → system suitability, peak id, integration, param calc, peak quality → confident qual+quant.

Use When

  • Review chrom data before reporting
  • Verify system suitability passes before sample seq
  • ID unknown peaks / confirm analytes by tR / spectral
  • Troubleshoot peaks, baseline anomalies, integration artifacts
  • Train analysts → interpret

In

Req

  • Chromatogram: Digital/printed w/ time + detector axes
  • Ref standard: tR + response of known analytes (same method)
  • Method params: Column, mobile phase/carrier, temp/gradient, detector

Opt

  • Spectral: UV-Vis (DAD), MS, or other → peak confirm
  • Prior chroms: Historical data (same method) → trend
  • Suitability criteria: Method / regulatory limits
  • Prep details: Dilution, recovery, IS conc

Do

Step 1: Verify System Suitability

Confirm system in spec before interpret.

ParameterTypical SpecificationCalculation
Retention time RSD<= 1.0%RSD of tR over n >= 5 injections
Peak area RSD<= 2.0% (assay), <= 5.0% (impurity)RSD of area over n >= 5 injections
Tailing factor (T)0.8-2.0 (USP), ideally 0.9-1.2T = W0.05 / (2 * f) where W0.05 = width at 5% height, f = front half-width
Resolution (Rs)>= 1.5 (baseline), >= 2.0 (regulated)Rs = 2(tR2 - tR1) / (w1 + w2)
Theoretical plates (N)Per column spec (e.g., >= 2000)N = 16(tR / w)^2 or N = 5.54(tR / w0.5)^2
Capacity factor (k')2.0-10.0 for primary analytek' = (tR - t0) / t0
  1. Locate suitability injections (usually 5-6 replicates ref std at seq start)
  2. Calc each param
  3. Compare vs acceptance criteria
  4. Any fail → system not suitable → don't proceed till fixed
  5. Doc results in batch record

→ All params in spec → system fit for purpose.

If err: tR RSD fails → check temp instability, mobile phase prep err, column degrade. Tailing fails → inspect inlet liner (GC) / column frit (HPLC). Res fails → test mix → replace column if needed.

Step 2: ID Peaks

  1. Compare peak tR vs ref std chrom
    • Acceptable: ±2% of ref tR (or ±0.1 min short runs)
  2. Ambiguous → co-injection (spike): add std to sample, re-inject. Target peak increases w/o broaden/shoulder.
  3. DAD HPLC: compare UV-Vis spectrum vs lib
    • Match index ≥ 990/1000 → positive ID
    • Check spectral purity across peak (front/apex/tail overlay)
  4. MS: confirm molecular ion (m/z) + key frag ions vs ref
  5. Flag unidentified → "unknown" w/ tR + rel response

→ All targets ID'd by tR match w/ spectral confirm where avail. Unknowns flagged w/ tR + area.

If err: tR uniformly shifted → systematic change (column age, temp drift, mobile phase err). Re-inject std → establish current tRs before re-eval.

Step 3: Integrate

  1. Select mode:
    • Auto w/ data sys defaults → start
    • Manual only when auto demonstrably misplaces baseline / peak boundary
  2. Set params:
    • Baseline sensitivity (slope / threshold)
    • Min area/height → reject noise
    • Peak width matches narrowest expected
  3. Verify baseline:
    • Connects start+end of peak at true baseline
    • Overlap → valley-to-valley / perpendicular drop per method
    • Gradient: rising baseline → tangent/exponential skim
  4. Check integration errs:
    • Split peaks as 2 when should be 1
    • Shoulder merged into main peak when should be sep
    • Noise spikes as peaks
    • Baseline thru peak (neg clip)
  5. Record final params + any manual + justification in audit trail

→ All targets integrated, correct baseline, no artifacts, manual docs w/ rationale.

If err: Auto consistently mishandles peak shape → timed-events integration w/ custom params for that window. Never adjust to achieve desired result → adjustments must be sci justified.

Step 4: Calc Chrom Params

Calc for all reported peaks:

  1. Resolution (Rs) adjacent:
    • Rs = 2(tR2 - tR1) / (w1 + w2)
    • Rs ≥ 1.5 → baseline sep; ≥ 2.0 → routine margin
  2. Tailing (T) at 5% height:
    • T = W0.05 / (2f)
    • 1.0 symmetric; >2.0 → significant tail
  3. Plates (N):
    • N = 16(tR / w)^2 baseline w, or N = 5.54(tR / w0.5)^2 half-h
    • Higher → better efficiency
  4. Capacity (k'):
    • k' = (tR - t0) / t0, t0 = dead time (void vol / flow)
    • Ideal 2-10 → good sep + reasonable run
  5. Selectivity (α) critical pair:
    • α = k'2 / k'1
    • α > 1.05 → adequate sep
  6. Tabulate all, compare vs method spec

→ All params calc, tabulated, compared vs criteria. Critical pair res + plate count documented.

If err: Plates below spec → column may be degraded → test fresh std, compare historical. Params drift in seq → investigate instrument stability.

Step 5: Assess Peak Quality

  1. Symmetry: Gaussian / near-Gaussian. Fronting (T < 0.8) → overload; tailing (T > 1.5) → secondary interactions / dead vol.
  2. Baseline sep: Quant → critical pairs must be baseline-resolved. Valley no return → note % valley + assess accuracy impact.
  3. Peak width consistency: Broader than expected (vs std) → on-column degradation, extra-column broadening, injection issues.
  4. Spectral purity (DAD/MS): Purity index → inhomogeneity → co-eluting impurity likely. Consider method adj for better res.
  5. Neg peaks / baseline disturb: Neg in UV → sample solvent absorbs more than mobile phase at λ → normal for solvent front, abnormal elsewhere.
  6. Ghost peaks: In blank → carryover, contaminated mobile phase, column bleed. ID source before report.
  7. Summarize quality + note limitations on reported results

→ Quality assessed per analyte. Anomalies (tail, co-elute, ghost) documented w/ data impact.

If err: Significant quality issues (co-elute confirmed by spectral impurity, ghost at analyte tR) → data may not be reportable. Flag, investigate root cause, re-run after corrective action.

Check

  • Suitability params calc + in spec
  • All targets ID'd by tR (± spectral)
  • Unknowns flagged w/ tR + area
  • Integration correct baseline, manual docs
  • Res, tail, plates, k' calc for all peaks
  • Quality assessed → no unresolved co-elute affecting quant
  • Ghost + carryover evaluated via blank
  • Results tabulated vs method acceptance

Traps

  • Accept auto integration w/o review: Data sys misplaces baselines, esp shoulders, small peaks near large, gradient baselines. Visual review always.
  • Confuse tR shift w/ new peak: Uniform tR shift (all move) → systematic change, not new compounds. Re-inject std → recalibrate before ID calls.
  • Report peaks below noise: S/N < 3 (detection) / < 10 (quant) → don't ID / quant. Calc S/N explicit for trace peaks.
  • Ignore solvent front: Void vol peak ≠ analyte. t0 correctly ID'd + excluded from reporting.
  • Manual integration → target result: Adjust to pass spec = data falsification. Changes must be sci justified + audit-trailed.
  • Neglect spectral purity: Clean peak can hide co-eluting impurity. Always check purity when DAD/MS avail.

  • develop-gc-method — method dev for GC producing chrom
  • develop-hplc-method — method dev for HPLC producing chrom
  • troubleshoot-separation — diagnose problems found during interpret
  • validate-analytical-method — formal validation of method generating data
  • interpret-mass-spectrum — detailed MS interpret for GC-MS / LC-MS peak confirm

GitHub リポジトリ

pjt222/agent-almanac
パス: i18n/caveman-ultra/skills/interpret-chromatogram
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

関連スキル

executing-plans

デザイン

executing-plansスキルは、完全な実装計画があり、それを管理されたバッチでレビューチェックポイントを設けながら実行する場合に使用します。このスキルは計画を読み込んで批判的にレビューした後、小さなバッチ(デフォルトは3タスク)でタスクを実行し、各バッチの間に進捗状況を報告してアーキテクトのレビューを受けます。これにより、品質管理チェックポイントが組み込まれた体系的な実装が保証されます。

スキルを見る

requesting-code-review

デザイン

このスキルは、コードレビュアーサブエージェントを起動し、処理を進める前に要件に対してコード変更を分析します。タスク完了後、主要な機能の実装後、またはmainブランチへのマージ前などに使用すべきです。このレビューは、現在の実装と元の計画を比較することで、問題を早期に発見するのに役立ちます。

スキルを見る

connect-mcp-server

デザイン

このスキルは、開発者がHTTP、stdio、またはSSEトランスポートを使用してMCPサーバーをClaude Codeに接続するための包括的なガイドを提供します。GitHub、Notion、カスタムAPIなどの外部サービスを統合するためのインストール、設定、認証、セキュリティについて解説しています。MCP統合のセットアップ、外部ツールの設定、またはClaudeのModel Context Protocolを扱う際にご利用ください。

スキルを見る

web-cli-teleport

デザイン

このスキルは、タスク分析に基づいて開発者がClaude Code WebとCLIインターフェースの選択を支援し、これらの環境間でのシームレスなセッションテレポーテーションを可能にします。Web、CLI、モバイル環境を切り替える際のセッション状態とコンテキストを管理することで、ワークフローを最適化します。様々な段階で異なるツールを必要とする複雑なプロジェクトにご活用ください。

スキルを見る