mramon_AERA2010

The use of R statistical software to analyze flow cytometry data

M. Ramón1, A. Maroto-Morales1, O. García-Álvarez2, P. Jimenez-Rabadán2, MD. Pérez-Guzmán2, F. Martínez-Pastor3, AJ. Soler1,

JJ. Garde1

1IREC, (CSIC-UCLM-JCCM), Albacete; 2CERSYRA, Valdepeñas; 3ITRA-ULE-INDEGSAL, León

10º Congreso de la Asociación Española de Reproducción AnimalCáceres, 2-5 Junio 2010

R Statistical SoftwareIntegrated suite of software facilities for data manipulation, calculation and graphical display.

WinMDI Version 2.9Windows Multiple Document Interface for Flow Cytometry.The Scripps Research Institute (TSRI)

WEASEL Version 2.7.4Walter & Eliza Analysis Software: Eclectic & Lucid. The Walter and Eliza Hall Institute (WEHI)

The use of R statistical software to analyze flow cytometry data10º Congreso de la Asociación Española de Reproducción Animal (AERA)Cáceres, 2-5 Junio 2010


Main Goal: To provide widespread access to a broad range of powerful statistical and graphical methods for the analysis of genomic data.

Bioconductor provides a unified framework to develop methods to analyze and interpret Flow Cytometry data.

Most common Packages

α FlowCore α FlowUtils

α FlowViz α FlowQ

α FlowStats α FlowClust

Bioconductor Open Source Software for Bioinformatics


A PRACTICAL EXAMPLE

• Semen sample evaluated at 2 different times (0 h and 3 h)

• Each semen sample was stained for:

• YO-PRO®-1. Apoptosis-like changes

• Propidium Iodide (PI). Membrane integrity

• MitoTracker® Deep Red FM (MT). Mitochondrial status


1. READ DATA FILES

> flowData <- read.flowSet( path =“.”, alter.names=TRUE, phenoData= “annotation.txt")

> wf <- workFlow(flowData, name="Workflow #1")


1. READ DATA FILES

> flowData <- read.flowSet( path =“.”, alter.names=TRUE, phenoData= “annotation.txt")

> wf <- workFlow(flowData, name="Workflow #1")

Organize standard flow cytometry data analysis in a workflow

• read.FCS()

• read.flowSet()

• list of flowSets

sampleId time stain

file.001 Y.PI.MT.0 0H FL1.H/FL3.H/SSC.W

file.002 Y.PI.MT.3 3H FL1.H/FL3.H/SSC.W

…

> wf

A flow cytometry workflow called 'Workflow #1'

The following data views are provided:

Basic view 'base view'

on a flowSet

not associated to a particular action item


2. STANDARD FLOW OPERATIONS

Set of basic operations common in Flow Cytometry Analysis

• Visualize Data

• Data Compensation

• Data Transformation

• Data Normalization

• Define Target Population (gating)



Scatter plot matrix



Scatter plot matrix

Some transformation is needed for better visualization of data

2a. DATA TRANSFORMATION

> tf <- transformList(colnames(flowData), asinh, transformationId="asinh")

> add(wf,tf)

> flowDataT <- transform(flowData, tf)


A new transformated flow data set


> tf <- transformList(colnames(flowData), asinh, transformationId="asinh")

> add(wf,tf)

> flowDataT <- transform(flowData, tf)


Assign new elements to an existing workflow

• linear transformation

• quadratic transformation

• log transformation

• asinh transformation

• …


Scatter plot matrix


2b. IDENTIFICATION OF SPERM POPULATION

FSC vs. SSC scatter plot

> xyplot(`SSC.H` ~ `FSC.H` | sampleId, data=wf [["asinh"]])




> xyplot(`SSC.H` ~ `FSC.H` | sampleId, data=wf [["asinh"]])


Debris



> sp.gate <- polygonGate(.gate=r01)

> add(wf, sp.gate, parent='asinh')





WinMDI WEASEL

3. MEMBRANE INTEGRITY STUDY

YO-PRO®-1 (FL1.H) vs. PI (FL3.H) scatter plot

> xyplot(`FL3.H` ~ `FL1.H` | sampleId, data=wf[["asinh"]])


RAW DATA





Debris must be removed !!!

RAW DATA





RAW DATA SUBSET SPERM POPULATION





Is a compensation of data needed?

Must be applied before data transformation !!

SUBSET SPERM POPULATION

2c. DATA COMPENSATION

COMPENSATION MATRIX

• Define our own compensation matrix.

• Provide on FSC files by the cytometer. Spillover() R Function

> cmat <- compensation(comp, parameters=colnames(Data(wf[["base view"]]))[3:6],

+ compensationId="comp")

> add(wf, cmat)

> flowDataC <- compensate(flowData, comp)


FL1.H FL2.H FL3.H FL4.H FL5.H

[1,] 1 0.240 0.032 0.001 0

[2,] 0.008 1 0.140 0.003 0

[3,] 0.170 0.170 1 0.210 0

[4,] 0.001 0.001 0.003 1 0

[5,] 0 0 0 0 1


COMPENSATION MATRIX

• Define our own compensation matrix.

• Provide on FSC files by the cytometer. Spillover() R Function

> cmat <- compensation(comp, parameters=colnames(Data(wf[["base view"]]))[3:7],

+ compensationId="comp")

> add(wf, cmat)

> flowDataC <- compensate(flowData, comp)



COMPENSATION MATRIX



COMPENSATION MATRIX


WEASEL


IDENTIFICATION OF DIFFERENT SUBPOPULATIONS

Definition of Regions/Quadrants


Simple geometric Filters• Rectangle Gate

• Polygon gate

• Quadrant Gate

Data-Driven Filters• KmeansGate

• Norm2Gate

• Curv1Gate and Curv2Gate

• …




RECTANGLE REGION POLYGON REGIONS QUADRANT REGIONS




QUADRANT REGIONS

-2 0 2 4 6 8 10

0.0

00

.10

0.2

00

.30

breakpoint for parameter FL1.H

N = 12668 Bandwidth = 0.329

De

nsi

ty

breakpointdens region

2 4 6 8

0.0

0.1

0.2

0.3

0.4

breakpoint for parameter FL3.H

N = 12668 Bandwidth = 0.1223

De

nsi

tybreakpointdens region



Definition of Regions/Quadrants


Regions are different for each semen sample

Data normalization will allow the use of a single region or quadrant set

2d. DATA NORMALIZATION

> norm <- normalization(parameters=param, normalizationId="norm",

+ normFunction=function(x, parameters, ...) warpSet(x,parameters))

> add(wf, norm, parent="asinh")


Y.PI.MT.0

Y.PI.MT.3

0 2 4 6 8 10

FL1.H

0 2 4 6 8 10

FL3.H

BEFORE NORMALIZATION

2d. DATA NORMALIZATION

> norm <- normalization(parameters=param, normalizationId="norm",

+ normFunction=function(x, parameters, ...) warpSet(x,parameters))

> add(wf, norm, parent="asinh")


Y.PI.MT.0

Y.PI.MT.3

-5 0 5 10

FL1.H

0 2 4 6 8 10

FL3.H

AFTER NORMALIZATION



Summarize statistics


FL1.H-FL3.H- FL1.H+FL3.H-

FL1.H+FL3.H+ Y.PI.MT.0 47.77% 33.68% 18.37% Y.PI.MT.3 53.65% 30.28% 15.15%

6. SUMMARY

• Read data

• Visualize data

• Compensation data

• Transform data

• Define Sperm population

• Normalize data

• Identify Subpopulations

• Get statistics



WHY TO USE R?

• Allows the analysis of several raw FCS files at once

• Use criteria base on data density to identify sperm subpopulations

• Definition of a unique set of filters/gates

• Reduction in time consumption

• Powerful statistical environmental

• Important development community

• Reproducibility research

7. iFLOW



THANK YOU FOR YOUR ATTENTION

mramon_AERA2010

Technology

flow cytometry data

data manipulation

use of r statistical

flow cytometry analysis

flow cytometry workflow

asociacin espaola

reproduccin animal cceres

reproduccin animal aera