---
title: "How to Choose a Metabolomics Provider"
lang: ko
url: https://www.gereallife.com/ko/services/resources-select-provider.html
canonical: https://www.gereallife.com/ko/services/resources-select-provider.html
entity: Shanghai Gereallife Biotech Co., Ltd
updated: 2026-11-02
---
KEY POINTS 
 
- Pick the technology from the question: untargeted to discover, targeted to validate, flux to explain mechanism
- Three QC essentials: QC samples, blanks and internal standards — missing any one is hard to defend in review
- Ask how identification was made: database match, authentic standard retention time, MS/MS fragments
- Ask for raw data: mzML/raw files and non-normalised peak areas, not only a normalised table 
 
 
 

## 011. Match the technology to your question

 

| Your goal | Appropriate approach | What to avoid |
| --- | --- | --- |
| Discover differential metabolites and generate hypotheses | Untargeted metabolomics (UHPLC-Q-TOF or Orbitrap full scan) | Starting with a targeted panel that only covers known analytes |
| Quantify 10–50 known analytes across groups | Targeted metabolomics (MRM with calibration curves) | Reporting relative peak areas as concentrations |
| Determine whether a pathway is genuinely switched on/off | Metabolic flux / isotope tracing (13C, 15N) | Static concentration changes alone |
| Broad screening across many metabolite classes | Widely targeted (large panel + MRM confirmation) | Single-class panels only |
| Lipid structures (acyl chains, double-bond position) | Lipidomics (optionally with CCS / ion mobility) | Routine reversed-phase untargeted only |

 
 
 
 

## 022. Judge QC, not the brochure

 

Every provider can show a tidy PCA plot. To judge data quality, ask these four questions directly:

 
- How many QC samples per batch, and what is the RSD distribution (target: more than 70% of features below 30% RSD)?
- Are blank and reagent-blank injections included to subtract background and detect contamination?
- Are internal standards isotope-labelled, and are they added before or after extraction?
- How long between sample preparation and acquisition, and are freeze–thaw cycles recorded? 
 
 
 

## 033. How confident is the identification

 

| Level | Evidence | Confidence | Typical use |
| --- | --- | --- | --- |
| Level 1 | Authentic standard: retention time and MS/MS match | Highest | Targeted quantification, biomarker confirmation |
| Level 2 | Library MS/MS match (GNPS, HMDB, MassBank) | Good | Annotating untargeted differences |
| Level 3 | Accurate mass and isotope pattern only | Low | Candidates only — not a conclusion |

 

Ask for the identification level of every reported metabolite. A report that gives names and scores without evidence cannot be used in a manuscript.

 
 
 

## 044. Deliverables to write into the contract

 
- Raw data: mzML or vendor raw files, mapped one-to-one to samples
- Non-normalised peak-area matrix plus the normalisation method and parameters
- QC report: RSD distribution, PCA, batch-effect assessment
- Identification table: name, m/z, retention time, adduct, identification level, match score
- Analysis code or pipeline version for reproducibility
- Biological interpretation: pathway enrichment (KEGG/HMDB) and statistics used 
 
 
 

## 055. How Gereallife works

 

Gereallife (Shanghai Gereallife Biotech Co., Ltd.) runs LC-MS and sequencing platforms for multi-omics services covering untargeted, targeted and widely targeted metabolomics, lipidomics, metabolic flux and isotope tracing, proteomics and sequencing. QC follows a QC-sample, blank and isotope-internal-standard workflow; raw data and non-normalised peak areas are delivered, and identification evidence is stated.

 
 
 

## Q&AFAQ

 Can untargeted and targeted analysis be combined? Yes, and it is usually the right design: discover with untargeted analysis on the same samples, then confirm the interesting metabolites by targeted MRM quantification. 
 How many samples do I need? It depends on effect size and variance. Cell or animal studies typically use at least 6–8 biological replicates per group; human studies require a power calculation based on the design, not a rule of thumb. 
 Can metabolomics establish mechanism? No. Metabolomics shows differences and associations; mechanism needs flux analysis, isotope tracing or functional experiments. Presenting correlation as causation is a common reviewer objection. 
 
 
 
 
 
 Need study design? 
 

Free study design and sample assessment, response within 1 hour.

 Contact the team → 
 
 
 
 

 
 Call us： 400-660-8826 
 Scan to chat on WeChat