1. The GOAL…
“That our differential expression data will
be an accurate representation of the
biological system measured, and that our
colleagues who develop clinical diagnostics
will find the data credible.”
2. The GOAL…
my BELIEF…
“That if I do not pay enough attention to
experimental design and quality control,
those colleagues will never believe me.”
“That our differential expression data will
be an accurate representation of the
biological system measured, and that our
colleagues who develop clinical diagnostics
will find the data credible.”
3. Variability in Label-Free Proteomics
Workflows
Sample Prep
Data Collection
Minimize
and
Measure
Variability!
4. Sample Prep: Procedure and Pitfalls
Lysis
Normalization
Digestion
SPE/Desalting
Dry and Reconstitute
Detergent/Chaotrope
Physical Stress
mL/mg volume ratios
Bradford/BCA/A280 Assay
Constant protein content
Constant volume
Substrate concentration
Constant Enz/Substrate Ratio
Fresh reagents
Mixing/Temperature/Time
Enzyme Quality?
Online/offline
Commercial/Homemade
Avoid where possible
Speedvac/Lyophilize
Solubility?
Avoid where possible
5. Sample Prep: Controlling for Variability
Lysis
Normalization
Digestion
SPE/Desalting
Dry and Reconstitute
Controlling for variability
Constant mass/volume ratios
Constant Volume and [Protein]
Protein Standards (SIL or “surrogate”)
Peptide Standards (SIL or “surrogate”)
6. How to Assess Variability of a Workflow
Zhang, Fenyo, and Neubert. J. Proteome
Res. 2009, 8(3): 1285-1292
Method 2 (Neubert et al)
𝑆2
𝑡𝑜𝑡𝑎𝑙
= 𝑆2
1
+ 𝑆2
2
+ 𝑆2
3
…+𝑆2
𝑛
Method 1 (‘old school’)
Biological
Technical
(Preparation)
Analytical
.
.
.
.
7. Limiting and Measuring Variability in LC-MS/MS
Analysis (Exemplar with n=6 Samples)
“Technical-Replicate Heavy”
Incorrect (Underestimation)
Correct
“Moderate Technical Replication” (“90/10 approach”)
“Singles + QC Pool”
Analysis Order:
1. Randomization (if unknowns)
2. “Blocking” + Randomization
(if known sample grouping)