[PPT] - Outline - IntroducAon into High-Content Screening (HCS) data and the PowerPoint Presentation

SLIDE 1

Screen Mining with KNIME

A user-friendly framework for high throughput / content data analysis

Mar$n Stöter HT - Technology Development Studio (TDS), the HC-Screening Unit at the MPI-CBG stoeter@mpi-cbg.de

KNIME workshop February 27th 2016, Berlin

SLIDE 2

Outline

Martin Stöter, MPI-CBG, Dresden, Germany

2

IntroducAon into High-Content Screening

(HCS) data and the HCS Tools nodes

Hands-on session HCS Tools
IntroducAon into ScripAng IntegraAon nodes
Hands-on session ScripAon IntegraAon

SLIDE 3

Technology Development Studio (TDS)

Martin Stöter, MPI-CBG, Dresden, Germany

3

MPI-CBG, Dresden, Germany Screening facility for academic laboratories Provide full service for automaAon and cell-based screens, RNAi and chemical screens Equipment: liquid handling robots, drop dispensers, plate washers, plate readers, High Content Screening plaTorms

SLIDE 4

Data Analysis is a Bottleneck in HCS!

4

Data analyst

Complex Experiments Lots of data (too much for Excel) Fancy data analysis / mining Many scienAsts, but few data analysts SomeAmes different languages Data analysis is oYen a boZleneck!

Scien$sts

HCS Tools

+ + + …

SLIDE 5

High-Content Screening (HCS) data

Martin Stöter, MPI-CBG, Dresden, Germany

5

Data generaAon

Cells (RNAi, compounds)
Microscopy -> images
Image analysis
Cell features/parameters -> well data

Tasks/problems

Read data from various sources

SQL database, XML, Excel, various .csv …

Screening specific staAsAcs
Screening specific uAliAes
Data mining, visualizaAon

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 A B DMSO DMSO DMSO C 0.001 DMSO DMSO 0.001 D 10 DMSO DMSO 10 E 10 DMSO DMSO 10 F 3 DMSO DMSO 3 G 3 DMSO DMSO 3 H 1 DMSO DMSO 1 I 1 DMSO DMSO 1 J 0.3 DMSO DMSO 0.3 K 0.3 DMSO DMSO 0.3 L 0.1 DMSO DMSO 0.1 M no AB no AB 0.1 DMSO DMSO 0.1 N no AB no AB 0.1 DMSO DMSO 0.1 O DMSO DMSO P

SLIDE 6

HCS Tools for KNIME

Data Import

Image Analysis Readers (Opera, OpereQa, Mo$onTracking) Plate Readers (Envision, GeniusPro, MSD SectorImager) Other (Example Data, Generic XML)

NormalizaAon

Percent-of-control (POC), Normalized percent inhibi$on (NPI) Z-score, B-score Vector Length Normaliza$on (clustering) Op$onal: robust sta$s$cs (Median + MAD) Select wells to normalize (controls, samples)

Quality Control

Z-prime factor (Z‘), Mul$variate Z‘, SSMD CV (coefficient of variance) Op$onal: robust sta$s$cs (Median + MAD) Select wells to normalize (controls, samples)

HCS Tools

SLIDE 7

HCS Tools for KNIME

UAliAes

Handle barcodes, wells and row leQers Join Layout from Excel (well annota$on, meta data) Create Well Posi$on (NEW)

VisualizaAon

Plate Heatmap Viewer Dose Response (dependent on R!)

Advanced StaAsAcs

BinningAnalysis

Data ManitupaAon / Pre-Processing

Split / Combine Colums (by header) Number FormaQer (NEW) Range Filter, SpliQer Outlier Removal

HCS Tools

SLIDE 8

HCS Tools: Standardized Data Format

Enforce standardizaAon of data format
Different reader nodes to shape a common data structure
Lower the knowledge entry barrier for new users

“barcode”, “plateRow”, “plateColumn”, param1, param2, …

> Eases up the usage of other HCS Tools nodes

SLIDE 9

HCS Tools: Expand well

StandardizaAon of the well coordinates:

“plateRow” and “plateColumn” as integer values resemble well posi$on matrix (instead of well)
Some nodes select these columns as default (Join Layout, Plate Heatmap Viewer)
Compa$ble with 96, 384 and 1536 well format
Plate Row Converter (leQer ↔ integer)
Create Well Posi$on (sortable well string)

NEW NODE

SLIDE 10

HCS Tools: Barcode Standard

Regular expression for interpretaAon of barcode:

Standardized table structure -> connec$on to our TDS compound database
(?<libplatenumber>[0-9]{3})(?<projectcode>[A-z]{2})(?<date>[0-9]{6})(?<replicate>[A-z]{1})
Configurable in Preferences -> KNIME -> HCA Tools
Mul$ple barcodes / regular expressions possible
Final release recently

SLIDE 11

HCS Tools: Barcode Standard

Regular expression for interpretaAon of barcode:

Standardized table structure -> connec$on to our TDS compound database
(?<libplatenumber>[0-9]{3})(?<projectcode>[A-z]{2})(?<date>[0-9]{6})(?<replicate>[A-z]{1})
Configurable in Preferences -> KNIME -> HCA Tools
Mul$ple barcodes / regular expressions possible
Final release recently

SLIDE 12

HCS Tools: Barcode Standard

Regular expression for interpretaAon of barcode:

Standardized table structure -> connec$on to our TDS compound database
(?<libplatenumber>[0-9]{3})(?<projectcode>[A-z]{2})(?<date>[0-9]{6})(?<replicate>[A-z]{1})
Configurable in Preferences -> KNIME -> HCA Tools
Mul$ple barcodes / regular expressions possible
Final release recently

SLIDE 13

HCS Tools: Annotate Experiment

Excel is the tool for experiment documentaAon and assay development Join Layout node is Excel Reader for defined spread sheet Plate format with mulAple well aZributes (1 plate layout -> 1 column in KNIME)

Title of layout starts in cell C5
Two empty rows between the layout

SLIDE 14

HCS Tools: Normalization

To compare data from different plates, days or runs data must be normalized per plate Selectable reference well populaAon per plate Percent-of-control (POC), Normalizes-percent-of- inhibiAon (NPI), Z-Score Robust staAsAcs (median & mad instead of mean & sd) with staAsAcs table as second output

SLIDE 15

HCS Tools: Normalization

To compare data from different plates, days or runs data must be normalized per plate Selectable reference well populaAon per plate Percent-of-control (POC), Normalizes-percent-of- inhibiAon (NPI), Z-Score Robust staAsAcs (median & mad instead of mean & sd) with staAsAcs table as second output

SLIDE 16

HCS Tools: Normalization

To compare data from different plates, days or runs data must be normalized per plate Selectable reference well populaAon per plate Percent-of-control (POC), Normalizes-percent-of- inhibiAon (NPI), Z-Score Robust staAsAcs (median & mad instead of mean & sd) with staAsAcs table as second output

SLIDE 17

HCS Tools: Quality Control (QC)

Quality control staAsAc measure the assay performance Selectable (mulAple) reference well populaAon per plate Z-Prime factor (Z’), mulAvariate Z’, strictly standardized mean difference (SSMD), coefficient of variance (CV) Robust staAsAcs (median & mad instead of mean & sd)

SLIDE 18

HCS Tools: Quality Control (QC)

Quality control staAsAc measure the assay performance Selectable (mulAple) reference well populaAon per plate Z-Prime factor (Z’), mulAvariate Z’, strictly standardized mean difference (SSMD), coefficient of variance (CV) Robust staAsAcs (median & mad instead of mean & sd)

SLIDE 19

HCS Tools: Quality Control (QC)

Quality control staAsAc measure the assay performance Selectable (mulAple) reference well populaAon per plate Z-Prime factor (Z’), mulAvariate Z’, strictly standardized mean difference (SSMD), coefficient of variance (CV) Robust staAsAcs (median & mad instead of mean & sd)

SLIDE 20

HCS Tools: Binning Analysis

Binning analysis describes changes in distribuAons Great tool for moving from cell to well data (instead of just taking mean per well)

"CellProfiler and KNIME: open source tools for high content screening.". Methods in molecular biology (Clifton, N.J.) 2013 986, S. 105-22

SLIDE 21

HCS Tools: Binning Analysis

Binning analysis describes changes in distribuAons Great tool for moving from cell to well data (instead of just taking mean per well)

"CellProfiler and KNIME: open source tools for high content screening.". Methods in molecular biology (Clifton, N.J.) 2013 986, S. 105-22

SLIDE 22

HCS Tools: Binning Analysis

Binning analysis describes changes in distribuAons Great tool for moving from cell to well data (instead of just taking mean per well)

"CellProfiler and KNIME: open source tools for high content screening.". Methods in molecular biology (Clifton, N.J.) 2013 986, S. 105-22

SLIDE 23

HCS Tools: Plate Viewer (discontinued)

Martin Stöter, MPI-CBG, Dresden, Germany

23

179 plates x 384wells = ~70.000 data points Ames x parameters

SLIDE 24