MS Preprocessing and Evaluation

1
MS Preprocessing and Evaluation

• Julien Prados

2
Introduction

• Context
• diagnosis and biomarker extraction from SELDI
  MALDI mass spectra
• Issues
• preprocessing mass spectra
• tune preprocessing parameters
• evaluate preprocessing quality

3
Work Flow

• Preprocessing
• Baseline Estimation
• MS Normalisation (TIC)
• Noise Estimation/Elimination
• Peak Detection
• Extract Peak Caracteristics
• Peak Alignment

Spectra with Control Diseased labels
List of Discriminant Features
A learning dataset
Classification model for prediction of patient
state from its mass spectrum
Machine Learning
4
Signal Distortions Correction

• Baseline estimated with open operator (local
  maxima of the local minima) in a sliding window
• Total Ion Current Normalization with baseline
  corrected signal and part of the signal gt 2000 Da

5
Noise Estimation / Elimination

• Noise estimated by standard deviation in a
  sliding window
• To determine what is a peak and what is not
• Possibility of signal smoothing (e.g. with
  wavelet, FFT)
• Better to work with raw data

6
Peak Detection
7
Extracting Peak Area

• p fixed point found by peak detection
• The signal is splitted into regions according to
  the minima between two consecutive peaks
• In each region, pl and pr are found by mean least
  square fitting of a piecewise linear model in two
  segments (horizontal and oblique)
• Area of the peak is given by area of the triangle
  (pl, p, pr)

8
Peak Alignement Missing Values

• Peak alignment performed by hierachical
  clustering (closest peaks are merged)
• Two strategies for missing values
• set missing values to zero because there is no
  peak
• retrieve signal intensity (not obvious for peak
  area)

?
?
?
9
Data Representation

• 3 possible data representations
• peak intensity signal intensity for missing
  values (is)
• peak intensity zero filling of missing values
  (iz)
• peak area zero filling of missing values (az)

10
Preprocessing Evaluation

• Solution 1 (Ideal case) Have samples with known
  content, or use a MS simulator and estimate peak
  detection performance.
• Solution 2 Do spectra replicates and estimate
  peak detection stability.
• Solution 3 In diagnostic applications, choose
  preprocessing parameters minimising
  generalisation error.

11
Choosing Peak Detection Parameter

• Compare detections between a normal MS spectrum
  and a blank one
• ? We can used a peak detection parameter of 2.5

12
Preprocessing Evaluation in Diagnostic (1/2)Data
Representation Evaluation

• is/iz ? filling missing values with signal
  intensity instead of zeroes retains more
  discriminatory informations
• iz/az ? using area or intensity does not result
  in significant differences
• is/raw ? no significant information lost in
  preprocessing, but a much more compact
  representation gain

13
Preprocessing Evaluation in Diagnostic
(2/2)Influence of Peak Detection Parameter

• Choosing is representation and SMO algorithm,
  what is the influence of peak detection parameter
  on the information content of the preprocessed
  datasets ?

14
Preprocessing Evaluation in Presence of
Replicates (1/2)

• perform peak detection and alignment

15
Preprocessing Evaluation in Presence of
Replicates (2/2)

• Estimate pourcentage of peaks find in at least
  10/10, 9/10, 8/10 ... 1/10 replicates

16
Conclusion

• Results depends heavily on parameter tunning, it
  should be done in an informed manner
• manual selection
• automatic selection
• We saw preprocessing pipeline of SELDI data.
  Preprocessing LC-MS data bring new issues
• More dimensions (2D, 3D)
• How to perform realtime peak detection and
  alignement of LCMS data ?
• How to perform realtime protein identification
  and guide MS/MS selection ?
• Challenge How to build learning methods able to
  deal with raw data

17
Peak Definition

• Valley definition for a point p The minimum
  points on left and right of p such that their is
  no point with intensity higher than the intensity
  of p between them.
• Peak definition a spectrum point p is considered
  a peak if its left and right valleys are deeper
  than the noise level.
• Remark
• No assumption on peak width

18
Data Representation Evaluation

• Error evaluation of 3 classification algorithms
• Instance Base Learning (IBk)
• Decision Tree (J48)
• Support Vector Machine (SMO)
• On 3 data representations
• peak intensity signal intensity for missing
  values (is)
• peak intensity zero filling of missing values
  (iz)
• peak area zero filling of missing values (az)
• For 3 datasets
• Stroke (Stk)
• Prostate Cancer (Pro)
• Ovarian Cancer (Ova)

19
Perspectives

• Preprocessing pipeline used in a reproducibility
  study of MALDI-TOF MS Zeferos et al. Sample
  Preparation and Bioinformatics in MALDI Profiling
  of Urinary Proteins. Submitted, JChromat, 2006
• Peak detection algorithm has been extended to 2D
  (and even nD) and applied on Nano-LC MS

20
2D Nano-LC Peak Detection
21
Thank you !
22
Preprocessing Objectives

• Correct signal distortions (baseline,
  normalization)
• Reduce dimensionality of learning problem (peak
  detection)
• Avoid removing discriminative informations

23
vd vs Error