Icon

AGM_​15Plates

Data Preprocessing Creates a new cell specific tag called 'Unique', eliminates columns with a large percentage of missing values,centersand scales the data based on the negative control, and filters out select columns with low variance and columns that arehighly correlative with one another. Data ReadingReads the object level CSVs (nuclei, cytoskeleton, and cells), joins them together, adds source file on columnname, and combines them with the metadata. Cross ValidationPartions the controls 80/20 such that the predictor predicts on just 20 percent of the controls. Right clickthe scorer node to see the accuracy of the model among other stats. The Cost Senstitive Classifierassigns costs to the controls such that false positive and false negatives weigh more than true positives/negatives. The highest cost is assigned to false negatives. Train the ModelUsing the same Cost Senstitive Classifier from the cross validation, but now the prediction isbased upon the whole data set. Scoring the ModelThis set of nodes segments the data on a per plate basis and subsequently scores each cell as being like the PC (taxol) or the NC (DMSO).These scores are then averaged on the per well basis and joined with the metadata. HitsIn these sets of nodes the hit list is determined. The hits are identified by calculating the mean + 3 sigma ofthe all the probabilities of being like the positive control. A row filter is then applied that cuts off all thecompounds with probabilities below that value. Right click the rule based row filter to see the hits beforewriting them to a CSV. A 6 sigma cutoff is also here under the 6 sigma row filter. NucleiNode 2Node 4CytoskeletonNode 6Node 7CellsNode 10Node 11Node 16Node 80Node 81Node 133Node 134Node 135Filter DataNode 519Node 588Node 589Node 604Node 656Node 657Node 658Node 659Node 660Plate,Well,Comp,ConcWellNode 668Node 669Node 420Node 419FinalNode 680Node 681Node 682Node 719Node 758Node 761MetadataNode 763Node 765Node 796Node 805UniqueNode 807Node 808Node 809Node 851Node 852Node 8533 SigmaNode 820 List Files Table Row ToVariable Loop Start Loop End List Files Loop End Table Row ToVariable Loop Start List Files Loop End Table Row ToVariable Loop Start CSV Reader CSV Reader CSV Reader Column Rename(Regex) Column Rename(Regex) Column Rename(Regex) Column Filter Normalize Plates(Z-Score) Partitioning Nominal ValueRow Filter Scorer Group Loop Start Cell Splitter Scorer Joiner Column Filter Column Filter GroupBy Sorter Column Filter Math Formula Math Formula Loop End Weka Predictor(3.7) CostSensitiveClassifier(3.7) Column Filter Weka Predictor(3.7) Joiner Joiner CSV Reader Column Filter Joiner Missing Value Missing ValueColumn Filter Column Combiner Low Variance Filter Linear Correlation(deprecated) Correlation Filter CSV Writer Rule-basedRow Filter Row Filter Math Formula Row Filter Data Preprocessing Creates a new cell specific tag called 'Unique', eliminates columns with a large percentage of missing values,centersand scales the data based on the negative control, and filters out select columns with low variance and columns that arehighly correlative with one another. Data ReadingReads the object level CSVs (nuclei, cytoskeleton, and cells), joins them together, adds source file on columnname, and combines them with the metadata. Cross ValidationPartions the controls 80/20 such that the predictor predicts on just 20 percent of the controls. Right clickthe scorer node to see the accuracy of the model among other stats. The Cost Senstitive Classifierassigns costs to the controls such that false positive and false negatives weigh more than true positives/negatives. The highest cost is assigned to false negatives. Train the ModelUsing the same Cost Senstitive Classifier from the cross validation, but now the prediction isbased upon the whole data set. Scoring the ModelThis set of nodes segments the data on a per plate basis and subsequently scores each cell as being like the PC (taxol) or the NC (DMSO).These scores are then averaged on the per well basis and joined with the metadata. HitsIn these sets of nodes the hit list is determined. The hits are identified by calculating the mean + 3 sigma ofthe all the probabilities of being like the positive control. A row filter is then applied that cuts off all thecompounds with probabilities below that value. Right click the rule based row filter to see the hits beforewriting them to a CSV. A 6 sigma cutoff is also here under the 6 sigma row filter. NucleiNode 2Node 4CytoskeletonNode 6Node 7CellsNode 10Node 11Node 16Node 80Node 81Node 133Node 134Node 135Filter DataNode 519Node 588Node 589Node 604Node 656Node 657Node 658Node 659Node 660Plate,Well,Comp,ConcWellNode 668Node 669Node 420Node 419FinalNode 680Node 681Node 682Node 719Node 758Node 761MetadataNode 763Node 765Node 796Node 805UniqueNode 807Node 808Node 809Node 851Node 852Node 8533 SigmaNode 820 List Files Table Row ToVariable Loop Start Loop End List Files Loop End Table Row ToVariable Loop Start List Files Loop End Table Row ToVariable Loop Start CSV Reader CSV Reader CSV Reader Column Rename(Regex) Column Rename(Regex) Column Rename(Regex) Column Filter Normalize Plates(Z-Score) Partitioning Nominal ValueRow Filter Scorer Group Loop Start Cell Splitter Scorer Joiner Column Filter Column Filter GroupBy Sorter Column Filter Math Formula Math Formula Loop End Weka Predictor(3.7) CostSensitiveClassifier(3.7) Column Filter Weka Predictor(3.7) Joiner Joiner CSV Reader Column Filter Joiner Missing Value Missing ValueColumn Filter Column Combiner Low Variance Filter Linear Correlation(deprecated) Correlation Filter CSV Writer Rule-basedRow Filter Row Filter Math Formula Row Filter

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
Created by: Reid Fursmidt
Created at: 2019-12-17
On NodePit since: 2021-11-16
Last update: 2024-04-18
Created with KNIME version: v4.3.2

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!