ScreenKNIMEing How HCS-Tools and Scripting Integrations can be used in a screening environment 1 Friday, February 24, 2012 1
Outline - 1 st half ‣ HCS-Tools step-by-step Setup / Preferences ‣ Todays task ‣ What we are working with ‣ The big goal ‣ Step by step ‣ 2 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 2
Outline - 2 nd half ‣ Scripting Integrations Setup / Preferences ‣ Node Types ‣ Configuration dialog elements ‣ Script editor ‣ Template repository ‣ ... ‣ Todays task ‣ 3 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 3
HCS-Tools 4 Friday, February 24, 2012 4
Setup / Preferences ‣ Installation http://tech.knime.org/update/community-contributions/nightly ‣ Community Contributions --> KNIME HCS Tools ‣ ‣ Preferences 5 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 5
Preferences ‣ Minimal samples required to calculate mean / median ‣ Minimal samples required to calculate variance / MAD ... at least x samples per group should be present to provide a ‣ descent estimate of the statistic important for normalization and QC nodes ‣ less samples will result in a warning ‣ ‣ Scaling factor for MAD set to the factor proposed for normal distributed data ‣ 6 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 6
Preferences ‣ Barcode patterns regular expressions (separated by ‘;’) which describes a barcode ‣ standard used for ‘Expand Barcode’ node to automatically retrieve meta ‣ data from the barcode, and ‘Plate Viewer’ node possible placeholders ‣ projectcode, libcode, libnumber, date, replicate, assay, ‣ description, concentration, concunit, timepoint, customa, customb, customc, customd ‣ (?<libplatenumber>[0-‑9]{3})(?<projectcode>[A-‑z]{2})(?<date> [0-‑9]{6})(?<replicate>[A-‑z]{1})-‑(?<libcode>[_A-‑z\d]{3})(? <assay>[-‑_\s\w\d]* 7 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 7
Todays task ‣ Where do we start? RNAi screen has been performed with a stable cell line. Cells ‣ were fixed and stained Nuclei and cytoplasm staining ‣ Marker 1 and Marker 2 ‣ 39 x 384-well plates ‣ 8 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 8
Todays task ‣ What are we working with? Results from an image analysis (Acapella) of images taken by the ‣ Opera - an automated confocal microscope from PerkinElmer Each RES-file (XML-format) contains the results of one 384 well ‣ plate Measurements ‣ Number of cells in the well ‣ several quality control measurements (intensities of different ‣ channels) The Layout is given as an Excel-sheet ‣ 9 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 9
The Layout ‣ An Excel-sheet of a certain format provides information on the treatment Positive Transfection Controls (Tox1, Tox, Tox3) ‣ Negative Controls (Mock, Untreated) ‣ RNAi library ‣ 10 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 10
Big goal ‣ Extract RNAi’s which are not toxic but show a significant increase of the signals in both marker channels 11 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 11
Step by step ‣ Load the raw data ‣ Add meta data from both barcode and layout ‣ Inspect data visually Strength of cell number reduction for transfection controls seems ‣ to be different, has to be quantified Readouts show a plate to plate variation ‣ ‣ Did the transfection work well? ‘well’ = at least 80% transfection efficiency ‣ 12 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 12
Step by step ‣ Load the raw data Data Input ‣ Add meta data from both barcode and layout ‣ Inspect data visually Strength of cell number reduction for transfection controls seems ‣ to be different, has to be quantified Readouts show a plate to plate variation ‣ ‣ Did the transfection work well? ‘well’ = at least 80% transfection efficiency ‣ 12 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 12
Step by step ‣ Load the raw data Data Input ‣ Add meta data from both barcode and layout Meta data integration ‣ Inspect data visually Strength of cell number reduction for transfection controls seems ‣ to be different, has to be quantified Readouts show a plate to plate variation ‣ ‣ Did the transfection work well? ‘well’ = at least 80% transfection efficiency ‣ 12 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 12
Step by step ‣ Load the raw data Data Input ‣ Add meta data from both barcode and layout Meta data integration ‣ Inspect data visually Visualization Strength of cell number reduction for transfection controls seems ‣ to be different, has to be quantified Readouts show a plate to plate variation ‣ ‣ Did the transfection work well? ‘well’ = at least 80% transfection efficiency ‣ 12 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 12
Step by step ‣ Load the raw data Data Input ‣ Add meta data from both barcode and layout Meta data integration ‣ Inspect data visually Visualization Strength of cell number reduction for transfection controls seems ‣ to be different, has to be quantified Readouts show a plate to plate variation ‣ ‣ Did the transfection work well? Quality Control ‘well’ = at least 80% transfection efficiency ‣ 12 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 12
Step by step ‣ Plate wise normalization / Percent of control (POC) plate data has to be normalized on the Mock wells ‣ two effects: ‣ Mock wells will be centered around 100% (eliminates plate ‣ wise variation) Transfection controls are represented as percentage which ‣ makes it easy to judge about transfection efficiency ‣ Quality control - SSMD Measure of how well positive control and negative control are ‣ separated from each other. It’s better interpretable than z prime factor 13 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 13
Step by step ‣ Plate wise normalization / Percent of control (POC) Normalization plate data has to be normalized on the Mock wells ‣ two effects: ‣ Mock wells will be centered around 100% (eliminates plate ‣ wise variation) Transfection controls are represented as percentage which ‣ makes it easy to judge about transfection efficiency ‣ Quality control - SSMD Measure of how well positive control and negative control are ‣ separated from each other. It’s better interpretable than z prime factor 13 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 13
Step by step ‣ Plate wise normalization / Percent of control (POC) Normalization plate data has to be normalized on the Mock wells ‣ two effects: ‣ Mock wells will be centered around 100% (eliminates plate ‣ wise variation) Transfection controls are represented as percentage which ‣ makes it easy to judge about transfection efficiency ‣ Quality control - SSMD Quality Control Measure of how well positive control and negative control are ‣ separated from each other. It’s better interpretable than z prime factor 13 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 13
Step by step ‣ Remove all screening plate where none of the transfection controls shows an efficiency < 80 % keep, if at least on transfection control has less then 20% cells ‣ ‣ Actual hit selection consider only wells with a cell number comparable to Mock ‣ (‘comparable’ = +- 0.5 standard deviation away from the median) consider wells which show more than 2 standard deviation ‣ increase of both marker channel signals Z-score normalization of the whole screen based on Mock ‣ centralize Mock values around 0 with standard deviation 1 ‣ 14 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 14
Step by step ‣ Remove all screening plate where none of the transfection controls shows an efficiency < 80 % keep, if at least on transfection control has less then 20% cells ‣ ‣ Actual hit selection consider only wells with a cell number comparable to Mock ‣ (‘comparable’ = +- 0.5 standard deviation away from the median) consider wells which show more than 2 standard deviation ‣ increase of both marker channel signals Normalization Z-score normalization of the whole screen based on Mock ‣ centralize Mock values around 0 with standard deviation 1 ‣ 14 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 14
Get ready ‣ Install the HCS-Tools (and Scripting Integrations) ‣ Download tutorial (Purpose) ‣ Where to find the node? ‣ How to configure the node? ‣ How it works? ‣ (What does the node view show?) ‣ What does the output table contain? ‣ ‣ Download workflow (including example data) ‣ 15 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 15
Scripting Integrations (R) 16 Friday, February 24, 2012 16
Setup / Preferences ‣ Installation http://tech.knime.org/update/community-contributions/nightly ‣ Community Contributions --> KNIME R Scripting Extension ‣ 17 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 17
Preferences 18 Antje Niederlein, niederle@mpi-cbg.de Friday, February 24, 2012 18
Recommend
More recommend
