From Vesicle Features to Cellular Phenotypes : Statistical - - PowerPoint PPT Presentation

▶
from vesicle features to cellular phenotypes statistical
SMART_READER_LITE
LIVE PREVIEW

From Vesicle Features to Cellular Phenotypes : Statistical - - PowerPoint PPT Presentation

Oct 18, 2023 209 likes •680 views

Introduction Mirko Birbaumer ECCB 2010 26th September 2010 1 / 25 From Vesicle Features to Cellular Phenotypes : Statistical Clustering in Image-based High-Throughput RNAi Screens Mirko Birbaumer birbaumer@imsb.biol.ethz.ch 26th September

slide-1
SLIDE 1

Introduction Mirko Birbaumer ECCB 2010 26th September 2010 1 / 25

From Vesicle Features to Cellular Phenotypes : Statistical Clustering in Image-based High-Throughput RNAi Screens

Mirko Birbaumer

birbaumer@imsb.biol.ethz.ch

26th September 2010

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-2
SLIDE 2

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 2 / 25 Motivation and Problem Description How a virus enters a cell and how it is transported within a cell.

RNA interference experiment : in each well there are thousands of (fixed) cells and the expression of a particular gene is silenced.

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-3
SLIDE 3

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 3 / 25 Motivation and Problem Description

Observed Vesicle Patterns in RNAi Screen

RNAi:Scrambled RNAi: MAPK7 RNAi: CDK8 RNAi: PRKAG1

2 1 3 4 2 1 3 4 3 1 2 4 3 1 2 5 4

A488-Tfn

C

How can we classify these patterns in an unsupervised manner?

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-4
SLIDE 4

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 4 / 25 Motivation and Problem Description

Motivation and Problem Description

In the context of RNAi screens, we have observed a plethora of different vesicle patterns

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-5
SLIDE 5

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 4 / 25 Motivation and Problem Description

Motivation and Problem Description

In the context of RNAi screens, we have observed a plethora of different vesicle patterns How can we quantify these patterns? How can these patterns be distinguished and classified in an unsupervised and automated manner?

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-6
SLIDE 6

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 4 / 25 Motivation and Problem Description

Motivation and Problem Description

In the context of RNAi screens, we have observed a plethora of different vesicle patterns How can we quantify these patterns? How can these patterns be distinguished and classified in an unsupervised and automated manner? Aim : Find functional modules of genes whose silencing leads to similar vesicle patterns reflecting the function of these genes

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-7
SLIDE 7

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 4 / 25 Motivation and Problem Description

Motivation and Problem Description

In the context of RNAi screens, we have observed a plethora of different vesicle patterns How can we quantify these patterns? How can these patterns be distinguished and classified in an unsupervised and automated manner? Aim : Find functional modules of genes whose silencing leads to similar vesicle patterns reflecting the function of these genes Furthermore, we want to reveal the regulatory structure between silenced genes in a RNAi screen based on vesicle features

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-8
SLIDE 8

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 5 / 25 Overview of Image Analysis Pipeline

Image Analysis Pipeline

Nucleus&cell segmentation SVM-filtering of mitotic, apoptotic & out-of-focus cells SE background subtraction&object detection; cell annotation Automated imaging at 40x 0.9 NA 138,240 images (9 sites/well; 3 channels; 3 z-planes for Tfn) 305,070 single cells at interphase & in-focus 20.8x106 single vesicles and endosomes SE-detected objects ( ) Imax 15’ AF488-Tfn CellTracer DAPI AF488-Tfn

2 3 4

3,072 wells (994 controls)

1 Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-9
SLIDE 9

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 6 / 25 Overview of Image Analysis Pipeline Vesicle Features

Vesicle Features

v9 v6 v7 v5 v8 v3 v2 l w v1 v4

Feature Description Type v1

Area Size

v2

Integrated intensity Cargo content

v3

Ellipticity Shape

v4

Radius containing 80% intensity Cargo concentration

v5

  • Rel. distance to nucleus

Subcellular position

v6

Number of neighbours within Local crowding

v7

Number of neighbours within Local crowding

v8

Radius containing 40%

  • f cell’s vesicles

Distance to

  • ther vesicles

v9

Radius containing 60%

  • f cell’s vesicles

Distance to

  • ther vesicles

c1

Number of vesicles per cell area Endocytic activity

c2

Clustering of spatial point pattern (Ripley’s K) Overall pattern

c3

Regularity of spatial point pattern (Ripley’s K) Overall pattern

single vesicle or endosome single cell

( )

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-10
SLIDE 10

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 7 / 25 Overview of Image Analysis Pipeline Nucleus Classification

Nucleus Classification

GUI for cell classification based on nuclei features and SVM (R package e1071); the GUI was written in python (Tkinter) in combination with rpy used as an interface to R

Reality in screens : presence of artefacts or cells are in different mitotic stages

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-11
SLIDE 11

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 7 / 25 Overview of Image Analysis Pipeline Nucleus Classification

Nucleus Classification

GUI for cell classification based on nuclei features and SVM (R package e1071); the GUI was written in python (Tkinter) in combination with rpy used as an interface to R

Reality in screens : presence of artefacts or cells are in different mitotic stages Aim: we want to keep only in-focus cells at interphase for the analysis CellProfiler provides 52 intensity, texture and shape features of the detected nuclei (feature extraction with CellProfiler)

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-12
SLIDE 12

Image Analysis Pipeline Mirko Birbaumer ECCB 2010 26th September 2010 7 / 25 Overview of Image Analysis Pipeline Nucleus Classification

Nucleus Classification

GUI for cell classification based on nuclei features and SVM (R package e1071); the GUI was written in python (Tkinter) in combination with rpy used as an interface to R

Reality in screens : presence of artefacts or cells are in different mitotic stages Aim: we want to keep only in-focus cells at interphase for the analysis CellProfiler provides 52 intensity, texture and shape features of the detected nuclei (feature extraction with CellProfiler) Nucleus Classification based on SVM

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-13
SLIDE 13

From Vesicle Features to Cellular Phenotypes Mirko Birbaumer ECCB 2010 26th September 2010 8 / 25 Clustering Approaches Clustering of Vesicle Patterns

RNAi:Scrambled RNAi: MAPK7 RNAi: CDK8 RNAi: PRKAG1

2 1 3 4 2 1 3 4 3 1 2 4 3 1 2 5 4

A488-Tfn

C

  • CNTRL

CNTRL CNTRL CNTRL

  • CDK8

CDK8 CDK8 CDK8 MAPK7 MAPK7 MAPK7 MAPK7

  • CNTRL

CNTRL CNTRL CNTRL

  • CDK8

CDK8 CDK8 CDK8 MAPK7 MAPK7 MAPK7 MAPK7

  • CNTRL

CNTRL CNTRL CNTRL

  • CDK8

CDK8 CDK8 CDK8 MAPK7 MAPK7 MAPK7 MAPK7

2 3 1 2 4 3 1 2 3 1 4 3 2 1 4 2 3 1 4 4 2 1 3 1 4 2 3 4 3 1 5 2 Measurements

Vesicle feature distributions per cell Vesicle-averaged single-cell vectors Mean values Dissimilarity (Euclidean) Vesicle feature distributions per cell Vesicle GMM per cell Kullback-Leibler divergence Vesicle feature distributions of all cells Vesicle GMM on all cells Dissimilarity (Euclidean) 7 vesicle subpopulations (see Fig. 3a)

Approach 1 Approach 2 Approach 3

Single-cell vectors from subpopulation fractions

  • 1,000
  • 500

500 1,000

  • 20
  • 10

10 20 30

  • 10
  • 5

5

  • 4
  • 2

2 4 5

  • 0.4
  • 0.2

0.0 0.2 0.4

  • 0.2
  • 0.1

0.0 0.1 0.2 0.3 0.4 2 2 1 4 4 1 1 2 3 2 1 4 3 3 3 4 5 PCA (1st 3 PCs) PCA (1st 3 PCs)

MDS MDS MDS

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-14
SLIDE 14

From Vesicle Features to Cellular Phenotypes Mirko Birbaumer ECCB 2010 26th September 2010 9 / 25 Clustering Approaches Vesicle Subpopulations

Vesicle Subpopulations

Biplots and spider graphs reveal properties of vesicle subpopulations.

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-15
SLIDE 15

From Vesicle Features to Cellular Phenotypes Mirko Birbaumer ECCB 2010 26th September 2010 10 / 25 Clustering Approaches Vesicle Subpopulations

Vesicle Subpopulations

Vesicles in Control cells are annotated with colored symbols corresponding to the related subpopulation.

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-16
SLIDE 16

Clustering of Gene Perturbations Mirko Birbaumer ECCB 2010 26th September 2010 11 / 25 Single Cell Clustering in Large Datasets

CDK8 PRKAG1 RET RET PRKAG1 MAPK7 MAPK7 PTK9 scrambled NEK7 STK25 STK25 MARK1 SKIP GFP STK21 GFP CDK8 CDK8 CDK8 WNK2 PRKAG1 PRKAG1 RPS6KA5 STK31 PIP5K1A PIP5K1A PRKY TEC STK40 SNF1LK5 MAP2K5

1 2 3 4 5 6 7 c1 c2 c3

Vesicle classes Cell Features

Pattern 1 Pattern 2 Pattern 3 Pattern 4 Pattern 5 Pattern 6 Pattern 7 Pattern 8

  • 2

2.5 3.5 2

5μm

Vesicle Patterns:

◮ To every cell among 305,646

cells in total a cell vector is assigned

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-17
SLIDE 17

Clustering of Gene Perturbations Mirko Birbaumer ECCB 2010 26th September 2010 11 / 25 Single Cell Clustering in Large Datasets

CDK8 PRKAG1 RET RET PRKAG1 MAPK7 MAPK7 PTK9 scrambled NEK7 STK25 STK25 MARK1 SKIP GFP STK21 GFP CDK8 CDK8 CDK8 WNK2 PRKAG1 PRKAG1 RPS6KA5 STK31 PIP5K1A PIP5K1A PRKY TEC STK40 SNF1LK5 MAP2K5

1 2 3 4 5 6 7 c1 c2 c3

Vesicle classes Cell Features

Pattern 1 Pattern 2 Pattern 3 Pattern 4 Pattern 5 Pattern 6 Pattern 7 Pattern 8

  • 2

2.5 3.5 2

5μm

Vesicle Patterns:

◮ To every cell among 305,646

cells in total a cell vector is assigned

◮ Partitioning around Medoids

reveals 8 distinct vesicle patterns.

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-18
SLIDE 18

Clustering of Gene Perturbations Mirko Birbaumer ECCB 2010 26th September 2010 12 / 25 Analysis of Movies

Vesicle Subpopulation Modeling Applied to Movies of GFP labeled Tfn

Movie of GFP labeled Tfn after Nocodazole was added. 63x Spinning Disk

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-19
SLIDE 19

Clustering of Gene Perturbations Mirko Birbaumer ECCB 2010 26th September 2010 13 / 25 Clustering of Gene Perturbations Clustering Based on Vesicle Pattern Abundance

Clustering Based on Vesicle Pattern Abundance

Module 5 Module 2 Module 6 Module 4 Module 1 Module 7 Module 8 Module 3 Module 9

Dissimilarity Matrix for 693 Genes with 9 Modules

Module 8 Module 3 Module 7 Module 6 Module 2 Module 5 Module 9 Module 4 Module 1 Pattern 4 Pattern 8 Pattern 7 Pattern 3 Pattern 5 Pattern 6 Pattern 2 Pattern 1

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-20
SLIDE 20

Clustering of Gene Perturbations Mirko Birbaumer ECCB 2010 26th September 2010 14 / 25 Clustering of Gene Perturbations Clustering Based on Vesicle Subpopulations

Clustering Based on Vesicle Subpopulations

Module 10 Module 9 Module 11 Module 1 Module 4 Module 5 Module 3 Module 7 Module 8 Module 6 Module 12 Module 2

Dissimilarity Matrix for 693 Genes with 12 Modules

−3 −2 −1 1 2 3

Cluster Modules Module 1 Module 4 Module 11 t_18 t_19 Module_12 t_20 t_23 t_30 t_22 t_29 t_26 t_24 t_28 t_21 t_25 Module_1 Module_11 t_17 t_14 t_13 t_15 t_16 Module_3 Module_5 t_27 Module_9 t_2 Module_10 t_3 t_5 t_1 t_9 t_11 t_10 t_12 t_8 Module_2 Module_8 5 10 15 20 25 Cluster Modules Module 2 Module 8

−3 −2 −1 1 2 3

Cluster Modules Module 3 Module 5 Cluster Modules Module 6 Module 7 Module 12

−3 −2 −1 1 2 3

Cluster Modules Module 9 Module 10

Vesicle and Cell Features

Normalized z−value

Vesicle and Cell Features

Vesicle Class 1 Vesicle Class 2 Vesicle Class 3 Vesicle Class 4 Vesicle Class 5 Vesicle Class 6 Vesicle Class 7 Intensity Regularity Clustering Vesicle Class 1 Vesicle Class 2 Vesicle Class 3 Vesicle Class 4 Vesicle Class 5 Vesicle Class 6 Vesicle Class 7 Intensity Regularity Clustering

−3 −2 −2 −1 −0.3 0.3 1 2 2 3

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-21
SLIDE 21

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 15 / 25 Motivation and Problem Description

NEM Applied to Image-Based RNAi Screen

Perturbing some genes may have an influence on a global process, while perturbing others affects subprocesses of it. Examples of this important concept: Signalling pathways or endocytic pathways. How do we establish a hierarchy in the observed perturbation effects recorded in our RNAi screen?

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-22
SLIDE 22

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 15 / 25 Motivation and Problem Description

NEM Applied to Image-Based RNAi Screen

Perturbing some genes may have an influence on a global process, while perturbing others affects subprocesses of it. Examples of this important concept: Signalling pathways or endocytic pathways. How do we establish a hierarchy in the observed perturbation effects recorded in our RNAi screen? Ansatz : Nested Effects Models (F.Markowetz 2005): Inference of a genetic hierarchy from the nested structure of observed perturbation effects

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-23
SLIDE 23

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 15 / 25 Motivation and Problem Description

NEM Applied to Image-Based RNAi Screen

Perturbing some genes may have an influence on a global process, while perturbing others affects subprocesses of it. Examples of this important concept: Signalling pathways or endocytic pathways. How do we establish a hierarchy in the observed perturbation effects recorded in our RNAi screen? Ansatz : Nested Effects Models (F.Markowetz 2005): Inference of a genetic hierarchy from the nested structure of observed perturbation effects Relevance in Endocytosis : there are proteins that regulate several endocytic pathways and other proteins are specific for one single pathway

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-24
SLIDE 24

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 16 / 25 Endocytic Pathways In our data Transferrin and IL2R which hijack different endocytic pathways were screened. Which genes regulate both pathways, which genes are specific for each pathway? Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-25
SLIDE 25

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 17 / 25 Application of NEM to RNAi Screen Data Data Matrix

Data matrix

Whereas discretization neglects the strength of effects, NEM with data matrix consisting of p-value densities allows a more subtle and sophisticated analysis (H. Froehlich 2007)

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-26
SLIDE 26

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 17 / 25 Application of NEM to RNAi Screen Data Data Matrix

Data matrix

Whereas discretization neglects the strength of effects, NEM with data matrix consisting of p-value densities allows a more subtle and sophisticated analysis (H. Froehlich 2007) Procedure to compute log-density matrix:

◮ (i) Compute p-value for every observation of a single feature under

nullhypothesis that feature values are distributed according to a t-distribution with mean and standard deviation estimated from the distribution of single feature values

◮ (ii) Fit a three-component beta-uniform mixture model (BUM) ◮ (iii) Result of BUM : log-density matrix: The log-densities can be

interpreted as log signal-to-noise ratios. A value > 0 means higher signal than noise, and a value < 0 a higher noise than signal.

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-27
SLIDE 27

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 17 / 25 Application of NEM to RNAi Screen Data Data Matrix

Data matrix

Whereas discretization neglects the strength of effects, NEM with data matrix consisting of p-value densities allows a more subtle and sophisticated analysis (H. Froehlich 2007) Procedure to compute log-density matrix:

◮ (i) Compute p-value for every observation of a single feature under

nullhypothesis that feature values are distributed according to a t-distribution with mean and standard deviation estimated from the distribution of single feature values

◮ (ii) Fit a three-component beta-uniform mixture model (BUM) ◮ (iii) Result of BUM : log-density matrix: The log-densities can be

interpreted as log signal-to-noise ratios. A value > 0 means higher signal than noise, and a value < 0 a higher noise than signal.

◮ Run NEM on log-density matrix Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-28
SLIDE 28

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 18 / 25 Application of NEM to RNAi Screen Data NEM Applied to Module Vectors

NEM Applied to Module Vectors

−3 −2 −2 −1 −0.3 0.3 1 2 2 3

Strong Effect Low Effect Module 1 Module 5 Module 7 Module 10 Module 11 Module 12 Module 6 Module 9 Module 8 Module 2 Module 4 Module 3 Regularity

  • Ves. Cl. 4
  • Ves. Cl. 7
  • Ves. Cl. 3

Intensity

  • Ves. Cl. 1
  • Ves. Cl. 6
  • Ves. Cl. 2

Clustering

  • Ves. Cl. 5

Module 3 Module 11 Module 1 Module 2 Module 10 Module 12

Module Vector: Median of all gene vectors within the same module. Module 3 contains most of Scrambled/GFP wells

Module 4

−3 −2 −2 −1 −0.3 0.3 1 2 2 3

Module 3 Module 5 Module 11 Module 1 Module 6 Module 2 Module 9 Module 10 Module 8 Module 7 Module 12

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-29
SLIDE 29

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 19 / 25 Application of NEM to RNAi Screen Data NEM Applied to STRING Modules

NEM Applied to STRING Modules

How do we find a set of kinases which constitutes a potential signalling pathway?

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-30
SLIDE 30

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 19 / 25 Application of NEM to RNAi Screen Data NEM Applied to STRING Modules

NEM Applied to STRING Modules

How do we find a set of kinases which constitutes a potential signalling pathway? Idea : String, a protein-protein interaction database Find k = 15, 20, 30 Kinome modules (with kmeans) from String database and run NEM on those modules (3 < numb. genes per module < 50)

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-31
SLIDE 31

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 19 / 25 Application of NEM to RNAi Screen Data NEM Applied to STRING Modules

NEM Applied to STRING Modules

How do we find a set of kinases which constitutes a potential signalling pathway? Idea : String, a protein-protein interaction database Find k = 15, 20, 30 Kinome modules (with kmeans) from String database and run NEM on those modules (3 < numb. genes per module < 50) Result: more than 80 networks based on String kinase modules!

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-32
SLIDE 32

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 20 / 25 Application of NEM to RNAi Screen Data Example of a NEM Based on a String Module

Example of a NEM Based on a String Module

Vesicle Class 7 Vesicle Class 3 Vesicle Class 5 Clustering (Rip.) Regularity (Rip.) Intensity Vesicle Class 1 Vesicle Class 6 Vesicle Class 2 Vesicle Class 4

MAPK13 CKMT1B AURKA EPHB2 ADCK5 MAP2K1IP1 MAP2K1IP PLK4 AURKAIP1 FLT4 MAGI2 NUAK1 LOC283155 FLJ13052 ALPK3 CDK7 AAK1 STK32C NTRK1 PFTK1 WEE1 ROR1 MAPK8 PRKY GAK DAPK3 GRK4 MGC4796 AKAP7 MAGI1 TTBK1 CIB3 AURKB SRPK2 STK10 WNK4 PAK7 PBK MGC40579 SHARPIN STK40 TSSK1 PRKAG1 PTK2B CHEK1 TLK1 ALS2CR7 UCK1 PKN3 AK5 −4 −3 −2 −1 −0.4 0.4 1 2 3 4 AURKAIP1 FLT4 MAGI1 MGC4796 GRK4 MAPK8 MAPK13 PRKAG1 PBK NTRK1 PRKY CHEK1 AKAP7 AURKB SRPK2 PFTK1 TLK1 STK10 PLK4 CKMT1B MAGI2 MAP2K1IP MAP2K1IP1 MGC40579 NUAK1 WEE1 EPHB2 WNK4 GAK DAPK3 PAK7 LOC283155 ALS2CR7 UCK1 FLJ13052 ROR1 SHARPIN ADCK5 AURKA ALPK3 TTBK1 CDK7 STK40 CIB3 PTK2B PKN3 AK5 STK32C TSSK1 AAK1

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-33
SLIDE 33

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 21 / 25 Characterization of Unknown Endocytic Pathways

Characterization of Unknown Endocytic Pathways

RNAi screen was carried out with 2 ligands : Transferrin and IL2Rβ. It is known that they take different endocytic pathways but little is known about the pathway taken by IL2Rβ

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-34
SLIDE 34

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 21 / 25 Characterization of Unknown Endocytic Pathways

Characterization of Unknown Endocytic Pathways

RNAi screen was carried out with 2 ligands : Transferrin and IL2Rβ. It is known that they take different endocytic pathways but little is known about the pathway taken by IL2Rβ All vesicles containing IL2Rβ are classified to 7 vesicle subpopulations

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-35
SLIDE 35

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 21 / 25 Characterization of Unknown Endocytic Pathways

Characterization of Unknown Endocytic Pathways

RNAi screen was carried out with 2 ligands : Transferrin and IL2Rβ. It is known that they take different endocytic pathways but little is known about the pathway taken by IL2Rβ All vesicles containing IL2Rβ are classified to 7 vesicle subpopulations To find global regulators for both pathways, we apply NEM to gene vectors with 20 effect reporters (10 for Tfn and 10 for IL2Rβ)

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-36
SLIDE 36

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 21 / 25 Characterization of Unknown Endocytic Pathways

Characterization of Unknown Endocytic Pathways

RNAi screen was carried out with 2 ligands : Transferrin and IL2Rβ. It is known that they take different endocytic pathways but little is known about the pathway taken by IL2Rβ All vesicles containing IL2Rβ are classified to 7 vesicle subpopulations To find global regulators for both pathways, we apply NEM to gene vectors with 20 effect reporters (10 for Tfn and 10 for IL2Rβ) By successively moving down the network topology, we expect a splitting between modules specific for Tfn pathway and modules specific for IL2Rβ pathway

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-37
SLIDE 37

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 22 / 25 Characterization of Unknown Endocytic Pathways Summary and Discussion

Summary and Discussion

NEM with morphological vesicle features as effect reporters is a new approach to reveal the regulatory structure of endocytic pathways! Possiblity to interprete the hierarchy of the regulatory network in terms of observable perturbations (effect of gene silencing on a subpopulation of vesicles!)

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-38
SLIDE 38

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 22 / 25 Characterization of Unknown Endocytic Pathways Summary and Discussion

Summary and Discussion

NEM with morphological vesicle features as effect reporters is a new approach to reveal the regulatory structure of endocytic pathways! Possiblity to interprete the hierarchy of the regulatory network in terms of observable perturbations (effect of gene silencing on a subpopulation of vesicles!) Critical Aspects

◮ Number of perturbed genes is much larger than the number of effect

reporters

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-39
SLIDE 39

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 22 / 25 Characterization of Unknown Endocytic Pathways Summary and Discussion

Summary and Discussion

NEM with morphological vesicle features as effect reporters is a new approach to reveal the regulatory structure of endocytic pathways! Possiblity to interprete the hierarchy of the regulatory network in terms of observable perturbations (effect of gene silencing on a subpopulation of vesicles!) Critical Aspects

◮ Number of perturbed genes is much larger than the number of effect

reporters

◮ Features are statistically not independent! Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-40
SLIDE 40

Nested Effects Models Mirko Birbaumer ECCB 2010 26th September 2010 22 / 25 Characterization of Unknown Endocytic Pathways Summary and Discussion

Summary and Discussion

NEM with morphological vesicle features as effect reporters is a new approach to reveal the regulatory structure of endocytic pathways! Possiblity to interprete the hierarchy of the regulatory network in terms of observable perturbations (effect of gene silencing on a subpopulation of vesicles!) Critical Aspects

◮ Number of perturbed genes is much larger than the number of effect

reporters

◮ Features are statistically not independent! ◮ Validitity of the found networks has to be shown either experimentally

  • r a well-known network should be at least reproduced

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-41
SLIDE 41

Agent-Based Model of Intracellular Transport Mirko Birbaumer ECCB 2010 26th September 2010 23 / 25 Agent-Based Modeling of Intracellular Transport

Simulation of Intracellular Transport

In-vitro experiments for validation of network hypotheses are time- and cost-intensive Alternative: in-sillico experiments with agent-based model of intracellular transport

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-42
SLIDE 42

Agent-Based Model of Intracellular Transport Mirko Birbaumer ECCB 2010 26th September 2010 23 / 25 Agent-Based Modeling of Intracellular Transport

Simulation of Intracellular Transport

In-vitro experiments for validation of network hypotheses are time- and cost-intensive Alternative: in-sillico experiments with agent-based model of intracellular transport We have reproduced with this model all 8 vesicle patterns previously described by means of computer simulations Free parameters of the model: transition rates between different biological states of vesicles which are controlled by protein concentrations Relation between silencing of a gene associated with a vesicle pattern and parameter values in model of intracellular transport leading to corresponding simulated pattern

Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-43
SLIDE 43

Agent-Based Model of Intracellular Transport Mirko Birbaumer ECCB 2010 26th September 2010 24 / 25 Implementation of Agent-Based Model of Intracellular Transport Vesicle patterns were reproduced with agent-based model of intracellular transport Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-44
SLIDE 44

Agent-Based Model of Intracellular Transport Mirko Birbaumer ECCB 2010 26th September 2010 25 / 25 Acknowledgment

RNAi Screen and Movies

◮ Lucas Pelkmans (ETHZ IMSB) ◮ Karin Mench (ETHZ IMSB) ◮ Prisca Liberali (ETHZ IMSB) ◮ Alexandre Grassart (Institut Pasteur) Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-45
SLIDE 45

Agent-Based Model of Intracellular Transport Mirko Birbaumer ECCB 2010 26th September 2010 25 / 25 Acknowledgment

RNAi Screen and Movies

◮ Lucas Pelkmans (ETHZ IMSB) ◮ Karin Mench (ETHZ IMSB) ◮ Prisca Liberali (ETHZ IMSB) ◮ Alexandre Grassart (Institut Pasteur)

Statistical Analysis

◮ Peter Buehlmann (ETHZ Sfs) ◮ Markus Kalisch (ETHZ Sfs) ◮ Holger Froehlich (University Bonn) Institute of Molecular Systems Biology www.imsb.biol.ethz.ch

slide-46
SLIDE 46

Agent-Based Model of Intracellular Transport Mirko Birbaumer ECCB 2010 26th September 2010 25 / 25 Acknowledgment

RNAi Screen and Movies

◮ Lucas Pelkmans (ETHZ IMSB) ◮ Karin Mench (ETHZ IMSB) ◮ Prisca Liberali (ETHZ IMSB) ◮ Alexandre Grassart (Institut Pasteur)

Statistical Analysis

◮ Peter Buehlmann (ETHZ Sfs) ◮ Markus Kalisch (ETHZ Sfs) ◮ Holger Froehlich (University Bonn)

Modeling of Intracellular Transport

◮ Frank Schweitzer (ETHZ MTEC) ◮ Jo Helmuth (ETHZ INF) Institute of Molecular Systems Biology www.imsb.biol.ethz.ch