Bioinformatics pipeline for revealing tumour heterogeneity Mustafa - - PowerPoint PPT Presentation

▶
bioinformatics pipeline for revealing tumour heterogeneity
SMART_READER_LITE
LIVE PREVIEW

Bioinformatics pipeline for revealing tumour heterogeneity Mustafa - - PowerPoint PPT Presentation

Aug 13, 2022 339 likes •602 views

Bioinformatics pipeline for revealing tumour heterogeneity Mustafa Anl Tuncel Department of Biosystems Science and Engineering Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anl Tuncel | 10.07.19 | 1 Mustafa

slide-1
SLIDE 1

| | Department of Biosystems Science and Engineering

Bioinformatics pipeline for revealing tumour heterogeneity

Mustafa Anıl Tuncel

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 1

slide-2
SLIDE 2

| | Department of Biosystems Science and Engineering

anilbey /in/aniltuncel anilbey

Mustafa Anıl Tuncel

Software Engineer @ ETH Zürich

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 2

Research interests § Data analysis workflows § Bioinformatics § Machine learning § Recommender systems

slide-3
SLIDE 3

| | Department of Biosystems Science and Engineering

Outline

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 3

§ Background

§ Biology prior § Single cell sequencing technologies § Mutations on DNA

§ DNA mutation trees

§ Tree model § MCMC moves

§ Pipeline

§ Snakemake § HDF5

slide-4
SLIDE 4

| | Department of Biosystems Science and Engineering

What is a cell?

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 4

Figure 1. Representation of cell, tissue, organ, system and organism. Retrieved from https://www.colscol.com/body-system/

slide-5
SLIDE 5

| | Department of Biosystems Science and Engineering

DNA from single cells

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 5

Figure 2. DNA structure. Retrieved from https://www.interleucina.org/

slide-6
SLIDE 6

| | Department of Biosystems Science and Engineering

§ Copy number variations

§ Deletion § Duplication

Structural mutations on DNA

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 6

§ Mutations from DNA of single cells § Heterogeneous § Have ancestors, children, siblings

slide-7
SLIDE 7

| | Department of Biosystems Science and Engineering

Trees to represent structural mutations

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 7

DNA:

Region 1 Region 2 Region 3 Region 5 Region 4

root

R1 : +1 R2 : -1, R3:+1 R4 : -1 R5 : +2 R1 : +1 R3 : -1

slide-8
SLIDE 8

| | Department of Biosystems Science and Engineering

Learning the tree

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 8 root

R1 : +1 R2 : -1, R3:+1 R4 : -1 R5 : +2 R1 : +1 R3 : -1

  • Dirichlet-multinomial model with overdispersion
  • We target maximising the tree posterior with an MCMC

scheme

  • Prune-reattach
  • Label swap
  • Add/remove events
  • Add/remove node
  • Condense/split node
  • Genotype preserving prune-reattach
slide-9
SLIDE 9

| | Department of Biosystems Science and Engineering

Prune-reattach

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 9 root

R1 : +1 R2 : -1, R3:+1 R4 : -1 R5 : +2 R1 : +1 R3 : -1

root

R1 : +1 R2 : -1, R3:+1 R4 : -1 R5 : +2 R1 : +1 R3 : -1

Before After

slide-10
SLIDE 10

| | Department of Biosystems Science and Engineering

Add / remove node

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 10 root

R1 : +1 R2 : -1, R3:+1 R4 : -1 R5 : +2 R1 : +1 R3 : -1

Before After

root

R1 : +1 R2 : -1, R3:+1 R4 : -1 R5 : +2 R1 : +1 R3 : -1 R1 : +1

slide-11
SLIDE 11

| | Department of Biosystems Science and Engineering

Condense / split node

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 11 root

R1 : +1 R2 : -1, R3:+1 R4 : -1 R5 : +2 R1 : +1 R3 : -1

Before After

root

R1 : +1 R2 : -1, R3:+1 R4 : -1 R5 : +2, R1:+1 R3 : -1

slide-12
SLIDE 12

| | Department of Biosystems Science and Engineering

Tree learned from mouse data

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 12

slide-13
SLIDE 13

| | Department of Biosystems Science and Engineering

What else is required?

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 13

§ Reproducibility in research § Scalability § Support for Multiple programming languages § Multi processing § Cluster execution § Resources management § Statistics about resource usages

slide-14
SLIDE 14

| | Department of Biosystems Science and Engineering

Workflow management system

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 14

slide-15
SLIDE 15

| | Department of Biosystems Science and Engineering

Snakemake

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 15

  • A Pythonic workflow management system
  • Extends the Python syntax
  • Follows the GNU make paradigm
  • Workflows are defined in terms of rules that define how to create output files from input

files

  • Dependencies between the rules are determined automatically
  • Benefits from Python libraries
  • Automated logging of the status
  • Suspend/resume workflow
  • A general-purpose workflow management system for any discipline
slide-16
SLIDE 16

| | Department of Biosystems Science and Engineering

Example: read mapping

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 16

slide-17
SLIDE 17

| | Department of Biosystems Science and Engineering

Example: read mapping (generalised)

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 17

slide-18
SLIDE 18

| | Department of Biosystems Science and Engineering 10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 18

DAG of jobs

slide-19
SLIDE 19

| | Department of Biosystems Science and Engineering 10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 19

Snakefile

slide-20
SLIDE 20

| | Department of Biosystems Science and Engineering 10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 20

Config file

slide-21
SLIDE 21

| | Department of Biosystems Science and Engineering 10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 21

Cluster execution

  • Configurable for LSF/BSUB scheduler
  • Allows scaling without changing the workflow
slide-22
SLIDE 22

| | Department of Biosystems Science and Engineering

HDF5

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 22

Data Metadata

Dataspace

3

Rank

Dim_2 = 5 Dim_1 = 4

Dimensions

Time = 32.4 Pressure = 987 Temp = 56

Attributes

Chunked Compressed Dim_3 = 7

Storage info

Integer

Datatype § Hierarchical data format v5 § Binary files

  • Easy to manage multiple datasets
  • Keeps metadata with data
  • Fast I/O operations & storage

space optimization (compressed binary files)

  • Platform/language independent
  • Self describing
  • No need to load whole data

HDF = Hierarchical Data Format

slide-23
SLIDE 23

| | Department of Biosystems Science and Engineering 10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 23

HDF5 wrappers in Python

h5py is a thin, pythonic wrapper around the HDF5

slide-24
SLIDE 24

| | Department of Biosystems Science and Engineering

Outline

10.07.19 Bioinformatics pipeline for revealing tumour heterogeneity. |. Mustafa Anıl Tuncel 24

§ Background

§ Biology prior § Single cell sequencing technologies § Mutations on DNA

§ DNA mutation trees

§ Tree model § MCMC moves

§ Pipeline

§ Snakemake § HDF5

Future work

  • Publish the method
  • Compare to clustering methods
  • Evaluate on simulated data
  • Show results on real data
  • Wrap up the workflow as a Python

package

  • Do the C++ bindings
  • Open source it
slide-25
SLIDE 25

Thank you!

anilbey /in/aniltuncel anilbey mtuncel@ethz.ch Mustafa Anıl Tuncel Software Engineer ETH Zurich