outline
play

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

Mar$n Stter KNIME workshop HT - Technology Development Studio (TDS), the HC-Screening Unit at the MPI-CBG February 27 th 2016, Berlin stoeter@mpi-cbg.de Screen Mining with KNIME A user-friendly framework for high throughput / content data


  1. Mar$n Stöter KNIME workshop HT - Technology Development Studio (TDS), the HC-Screening Unit at the MPI-CBG February 27 th 2016, Berlin stoeter@mpi-cbg.de Screen Mining with KNIME A user-friendly framework for high throughput / content data analysis

  2. Outline - 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 2 Martin Stöter, MPI-CBG, Dresden, Germany

  3. Technology Development Studio (TDS) 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 3 Martin Stöter, MPI-CBG, Dresden, Germany

  4. Data Analysis is a Bottleneck in HCS! Complex Experiments Lots of data (too much for Excel) Fancy data analysis / mining Scien$sts Many scienAsts, but few data analysts SomeAmes different languages Data analysis is oYen a boZleneck! HCS + + Tools Data analyst + … 4

  5. High-Content Screening (HCS) data 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 5 Martin Stöter, MPI-CBG, Dresden, Germany P

  6. HCS HCS Tools for KNIME Tools 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)

  7. HCS HCS Tools for KNIME Tools 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

  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

  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

  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

  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

  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

  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

  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

  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

  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

  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)

  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)

  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)

  20. HCS Tools: Binning Analysis "CellProfiler and KNIME: open source tools for high content screening.". Methods in molecular biology (Clifton, N.J.) 2013 986, S. 105-22 Binning analysis describes changes in distribuAons Great tool for moving from cell to well data (instead of just taking mean per well)

  21. HCS Tools: Binning Analysis "CellProfiler and KNIME: open source tools for high content screening.". Methods in molecular biology (Clifton, N.J.) 2013 986, S. 105-22 Binning analysis describes changes in distribuAons Great tool for moving from cell to well data (instead of just taking mean per well)

  22. HCS Tools: Binning Analysis "CellProfiler and KNIME: open source tools for high content screening.". Methods in molecular biology (Clifton, N.J.) 2013 986, S. 105-22 Binning analysis describes changes in distribuAons Great tool for moving from cell to well data (instead of just taking mean per well)

  23. HCS Tools: Plate Viewer (discontinued) 179 plates x 384wells = ~70.000 data points Ames x parameters 23 Martin Stöter, MPI-CBG, Dresden, Germany

  24. HCS Tools: Plate Heatmap Viewer VisualizaAon of screening campaigns with meta data Easy to find visually paZers, driYs, errors… New features: - KNIME Colors - HiLite support - representaAon of images - many different configuraAons, e.g. color scale…

  25. HCS Tools: Plate Heatmap Viewer VisualizaAon of screening campaigns with meta data Easy to find visually paZers, driYs, errors… New features: - KNIME Colors - HiLite support - representaAon of images - many different configuraAons, e.g. color scale…

  26. HCS Tools: Plate Heatmap Viewer VisualizaAon of screening campaigns with meta data Easy to find visually paZers, driYs, errors… New features: - KNIME Colors - 10 x 384well plate - HiLite support - 3 replicates - ~10,000 data points - representaAon of images - Raw data - many different configuraAons, e.g. color scale… - Meta data from barcode - Normalized data - Different readout - Meta data from layout - Browsing single plate - Viewing the well data - Display of images - … more

Recommend


More recommend