SLIDE 1
Signal Conditioning and Filtering
- f SELDI MS Time Series
Sij
Sij
LONG WAY!
100 µm
SELDI Profiling: Sources of Error
D
Signal Conditioning and Filtering of SELDI MS Time Series Dariya - - PowerPoint PPT Presentation
Signal Conditioning and Filtering of SELDI MS Time Series Dariya - - PowerPoint PPT Presentation
Signal Conditioning and Filtering of SELDI MS Time Series Dariya Malyarenko College of William and Mary & INCOGEN, Inc. new data familiar problems known techniques SELDI Profiling: Sources of Error S ij D S ij 100 m LONG WAY! Signal
SLIDE 2
SLIDE 3
Signal Conditioning Issues
- Peak relevance: biological or artifact?
- noise characterization
- Peak location: which column entry?
- deconvolution filtering
- Peak intensity: which value entry?
- background removal
Sij
SLIDE 4
Baseline Correction
Peak intensity in the data matrix
Baseline is predictable from RC modeling Single shot or average data
SLIDE 5
Nonlinear Baseline
Peak intensity and biological relevance
Length is proportional to the number of overload points Single shot correction only
SLIDE 6
Baseline Ringing
Peak relevance
2000 0.2 0.1 4000 0.0
CHIRP
Time Frequency Time Intensity
Damped coherent oscillation after overload Should be subtracted after each single shot
SLIDE 7
Smoothing & Rescaling
Peak relevance and intensity
MH+ MH2+
SLIDE 8
Rescaling & Noise
Peak relevance
Provides constant noise amplitude Filtered noise is “colored”
SLIDE 9
Subposition De-Jittering
Peak location, intensity, relevance
Subposition de-jittering during acquisition
SLIDE 10
SELDI Resolution in Time
Peak size and location
σt = const (< 12 kDa): monoisotopic target (Na) can be used in time
SLIDE 11
Shaping and Spiking Target Filter
shaping denoising
W
- 1
deconvolution
W
Peak location, intensity
SLIDE 12
Deconvolution of SELDI TOF
Peak location, intensity, relevance
+Na
- H2O
- CO2
MH++SA M/z
TARGET Na & SA adducts and H2O, NH4 & CO2 (neutral loss) peaks are detected up to 9 kDa Adduct peaks are correlated
SLIDE 13
M/z, Da
Spiking Accuracy
Target filter preserves total intensity Experimental filter detects peak shifts < σt in simulated data
Peak location and size
SLIDE 14
Conclusions:
Noise characterization and reduction:
- baseline removal
- detector saturation correction
- SNR rescaling
Location calibration:
- correlative dejittering
Resolution enhancement;
- target filter deconvolution
SLIDE 15