# Sds Gel Review > Review SDS-PAGE or protein purification gel images using DNA sequence, protein sequence, base-pair length, expected protein size, and lane labels. Use when the user wants to judge whether a gel ran well, whether the main band matches the expected product, or whether there may be impurities, degradation, aggregation, or low expression. - Skill: `runchuan-bu/sds-gel-review` (Agent Skill) - Install (CLI): `npx skillmds@latest add runchuan-bu/sds-gel-review` - Raw SKILL.md: https://api.skillmd.com/api/skills/runchuan-bu/sds-gel-review/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: Runchuan-BU (https://skillmd.com/u/runchuan-bu) - Updated: 2026-09-09 - Page: https://skillmd.com/skills/runchuan-bu/sds-gel-review --- # SDS-Gel Review Use this skill when the user provides a gel image plus any of: - DNA sequence - protein sequence - coding sequence length in bp - expected protein size in kDa - tag, construct, host, purification step, or lane labels This is an interpretation skill, not a definitive assay. Distinguish what is directly observed from what is inferred. ## Main goal Given a gel image and partial sequence/context, produce a practical lab-style judgment: - did the gel run cleanly or poorly - is there a plausible main band - does the main band roughly match the expected target - are there obvious impurities, degradation, smearing, or aggregation - what extra information would most improve confidence ## Inputs to request or infer Collect as many of these as possible before judging: - gel image - whether the image is SDS-PAGE, western blot, or native gel - DNA sequence or protein sequence - if only bp is given, whether it is coding sequence length - expected protein name - expected molecular weight in kDa - expression host - tag or fusion partner - purification step or lane meaning - whether reducing conditions were used - whether a ladder is visible If some are missing, continue with a lower-confidence interpretation. ## How to reason ### 1. Normalize the sequence information - If the user provides a protein sequence, use it directly. - If the user provides a DNA sequence, treat it as coding sequence only if the context supports that. - If the user provides only bp length, estimate protein length as `bp / 3` aa only when it is likely a coding region. - Estimate theoretical protein mass with the rough rule `110 Da per amino acid`. - Convert to kDa and note that tags, signal peptides, cleavage, glycosylation, oligomerization, and unusual composition may shift apparent migration. ### 2. Read the gel image conservatively Inspect the image for: - lane count and lane boundaries - visible labels above or below lanes - presence or absence of ladder - strongest band in each relevant lane - approximate relative migration of the strongest band - extra lower bands suggesting degradation - extra upper bands suggesting dimers, aggregates, uncleaved fusion, or contaminants - smearing, overloaded lanes, distorted fronts, or uneven running - enrichment or loss across purification lanes if the labels indicate flow-through, wash, elution, pellet, or lysate If the ladder is missing, use labeled lane order and relative migration only. Do not invent exact kDa values. ### 3. Compare expectation vs observation Decide whether the image most supports one of these: - likely target band at approximately expected size - possible target band but confidence limited - target band not clearly visible - strong expression but poor purity - purified sample but with degradation - aggregation or high-MW species likely - gel quality too poor for a reliable call ### 4. Explain confidence Confidence is higher when: - ladder is visible - expected protein size is provided - lane labels are clear - construct/tag information is available - the gel has multiple purification steps that tell a consistent story Confidence is lower when: - only bp count is provided - ladder is absent - lane labels are missing - image quality is low - the construct may include tags or cleavage not described ## Important guardrails - Never claim a precise molecular weight from the image when no ladder is visible. - Never claim a band is definitely the target protein from DNA length alone. - Clearly separate: - Observation: what the image visibly shows - Inference: what might explain it - If the image quality or metadata are insufficient, say so plainly. - Prefer practical lab wording over overconfident structural biology language. ## Output format Use this structure: ### Overall judgment - One short paragraph on whether the gel looks interpretable and whether the target seems present. ### What I can directly see - lane pattern - strongest band(s) - impurities, smear, aggregation, or degradation - whether purification appears to improve the sample ### Expected size estimate - from protein sequence, DNA sequence, or bp count - state assumptions clearly ### Does the observed band roughly match? - yes / maybe / unclear / no - explain why ### Main concerns - 1 to 3 likely issues only ### What would help next - ask for the smallest missing item that would improve confidence, such as: - ladder annotation - lane labels - exact coding sequence length - tag/fusion information - expected kDa - original uncropped image ## Practical heuristics - Approximate aa count from coding sequence: `bp / 3` - Rough protein mass: `aa * 110 Da` - His-tag or small peptide tags usually shift size only slightly - Large fusion tags can meaningfully shift migration - Smear below the main band often suggests degradation or proteolysis - Signal near the top or stacking boundary can suggest aggregation or incomplete denaturation - A very thick single band with little else may still reflect overloading ## Example task types - "Here is a DNA sequence and SDS-PAGE image. Did my purification work?" - "This construct is 978 bp. Does the main band in this gel make sense?" - "I expect a 42 kDa protein. Which lane looks best?" - "No ladder on this gel, but the lanes are labeled lysate, wash, elution. Please judge whether the elution looks clean."