This demo will show the analysis functionality of Phenom-Networks based on a dataset generated in the Hebrew University, the Faculty of Agriculture in.

This demo will show the analysis functionality of Phenom-Networks based on a dataset generated in the Hebrew University, the Faculty of Agriculture in Dani Zamir’s laboratory with additional data generated at Cold Spring Harbor in Zachary Lippman’s laboratory. All the data presented here has been prepared for publication in Krieger et al, (2009). To initiate the tutorial you will need to select the “Tomatoes” database and log in the system either as a guest or as a registered user. .

Content

• Slides 3-23: General demonstration of some of the analysis capabilities of Phenom Networks.

• Slides 24-33: Recipes to generate an equivalent analysis as in Figure 1 (B and C) of Krieger et al. 2009 (Submitted).

After you log in, go to “Phenotype -> analysis”. This is the main analysis page of the system which allows the user to perform statistical analysis of the data. The analysis options appear on the left side, organized into categories like “univariate”, “multivariate” and so on. In the middle, there is a table that lists all traits and on the right there are fields (which depend on the selected analysis) that will contain the traits that will be analyzed. The toolbar includes filtering possibilties to subset the data as well as other options.

Phenotype -> analysis

The first step before starting to analyze is to select the study or studies that you are interested in. For this click on the “studies filter” button from the toolbar. This will open a popup window that contain all available studies. Let’s select the “mutant ODO 09 Akko” study and click on OK.

Studies filter button

After you click OK, the popup window disappears, the icon on the “studies filter” button is changed to filtered mode and the selected study is indicated on the status bar (the green line on the bottom). The trait’s list now presents only the traits that were measured in the selected study. Now we are ready to invoke the analyses.

Studies filter changed to filter mode

The selected study is indicated in the status bar

Let’s select the “Fit Y by X” option (this follows JMP terminology. See: http://www.jmp.com/about/ ). In order to define the analysis, we need to select the traits and put them in the appropriate fields (on the right). We can think of the traits as columns of a table and by assigning them to roles (X or Y variables in this case) we define the analysis. Lets select for example the traits “Total yield” and “Brix”. After they are selected (you can use the Ctrl button to select them both) click on the “Y. variables” to move them to the field. Then select the “GERMPLASM IDENTIFICATION (breeder name)” and put it on the “X variables” field. This is the trait that defines the genotype (This terminology was defined by the researcher when he uploaded the data). finally click “OK”.

1) Select “Fit Y by X” 2) Select “Brix” + “Total yield” and click on the “Y variables” button

3) Select “GERMPLASM IDENTIFICATION – breeder name” and click on the “X variables” button.

4) Click “OK”

The result of the analysis is shown on a separate tab - “Result 1” in this case. It presents two figures: one for total yield and another for brix. Similarly, I could select more traits as “Y variables” and generate more figures – one for each trait.

Explanation about the figure: the X axis shows all the levels of the trait GERMPLAS IDENTIFICATION (that I selected as “X variables”), which are the genotype’s names. The data points above each level represent all observation units of the particular genotype and their Y axis shows the measurements of the trait, which is indicated on the figure’s title (“Total yield” for the first figure). The diamonds middle horizontal line indicate the genotype’s mean, and the diamond’s length is the confidence interval. Pointing the mouse on each data point will show the observation unit ID and its value. The “Show table” button will display the table used for this calculation with the possibility to download it to Excel.

Save analysis – the output of each analysis can be saved by clicking on the “Save” button. A popup window will appear showing previously stored analyses (if any). you can then give a name to the current analysis, select the folder and click “Save”.

1) Click on “Save”

2) The popup window show previous analyses (if any). Give a name and click “Save”.

Only registered users can use this functionality and save analyses. Guests are not allowed to do this.

We can go back to the “Form” tab. Now I’ll show an example of multivariate analysis. For this let’s pick the “bars” option under “Multivariate” (from the left panel). Note that the traits we classified before stayed in their fields, so we can just click the “OK” button.

In Multivariate analyses, a single figure is generated that shows all traits in it (in contrast to Univariate where it produces a figure for each trait). Here, the zero line represents the mean value for each trait and the bars show the genotype’s value for each trait as percent of the total mean. We can generate this figure such that the zero line will represent one of genotype’s mean instead of the total mean (I’ll show that later). You can point the mouse on each bar to see its corresponding genotype and trait. You can redo this analysis by picking more than two traits. It will show them all.

let’s go back to the “Form” tab, select another multivariate analysis - “means heatmap”, then select all variates and move them to the “Y variables” field. Then click OK.

The output shows a heatmap where all genotypes are on the Y axis and all traits are in the X axis. Each little colored rectangle indicates the mean value of the corresponding genotype in the corresponding trait; green color indicate low value, red is high and black is average (see the color scale on the top left of the figure). You can point the mouse on each such rectangle to see its corresponding genotype and trait.

Now I’ll redo the “Fit Y by X” analysis of the trait “Total yield”, but this time I’ll split the output according to additional factor – “hormones spray”. In order to test the effect of the hormone treatment, some of the plants were treated with an anti-Gibberellic acid hormone spray in the nursery, and some were not. This analysis will generate two figures: the first includes only treated plants and the second non treated plants. For doing this I go to the “split by” section on the rightmost side and choose the “Factor” option. Another field will appear (“By”) below the “X variables”. in this field I put the “hormones spray” factor. Then press OK.

1) Split by factor

2) Add “hormones spray to the “By” field

Here are the results: two figures were generated for “total yield”: the first for plants that were not treated with hormones (indicated on the figure’s title as “hormones spray of level: 0”), and the second for plants that treated with hormones.

In this analysis I want to include only some of the genotypes that were included in the study, instead of doing it over all of them. For this I click on the “factors filter” button in the toolbar (the second button). A popup window appears that display on its top a list of all available factors. As a first step I select the “GERMPLASM IDENTIFACTION – breeder name” factor and then under “Conditions” I choose the “in” operator. This will load all genotype names to the “labels” table, from which I can select the desired genotypes and add them to the conditions list using the button “Add condition”. Then click on “OK”. This filtering functionality is equivalent to selecting of particular rows from the table.

Factors filter

This is how the page looks like after I finished the filtering. The “factors filter” button was changed to filtered mode, and the factor that was filtered is indicated on the status bar (bottom). Now I can click “OK”.

Factors filter changed to filtered mode

Indiacation in the status bar

Now the figure shows only the selected genotypes. When you click the button “Show table” (on the bottom), three tables will appear and the button will change to “Hide table”.

Show / Hide tables

I can perform means comparison by clicking the “Compare means” link (on the right) that invoke a new window. In the window I select the statistical test (Students t, Dunnetts or Tukey), then the “a” threshold (defaults to be 0.05), and finally I choose the genotype (level) to compare to.

1) Compare means statistically

2) Select Dunnett

3) Set alpha threshold

4) Select control

The output displays the “M82” genotype in red bold that indicates that this is the level that all other genotypes are compared to. Then, “sft-stop” is in light-red, indicating that it is not significantly different than M82 under the selected a threshold (0.05). All other genotypes are black, meaning they are significantly different from M82.

If we supply the “GERMPLASM IDENTIFICATION – SFT combined” to the “X variables” field (instead of “breeder name”), it’ll combine all alleles of sft into a single genotype.

We select “SFT combined” instead of “breeder name”, and put it on the “X variables”.

All SFT alleles are combined together. Thus sft-4537, sft-7187 and sft-stop are all named sft. Similarly, sft-4537 x M82, sft-7187 x M82 and sft-stop x M82 are all under “sft x M82”. The statistical analysis shows all genotypes are significantly different than M82 (as they are written in black and not in red).

Here I select two experiments (in Akko and in Massarik).

In case you choose “split by: study”, the system will output a figure for each study. If you choose “split by: none”, all studies will be combined together in a single figure

This is the result of the two studies that are combined into a single figure. The red points represent the replications from the first study and the blue from the second one.

Krieger et al (unpub): Figure 1

The next slides will explain how to produce a figure that is equivalent to figure 1 of Krieger et al.

Click on the “Studies filter” button and select the appropriate study: “mutant ODO 09 Akko”.

1) Click on the “studies filter” button to open the studies window.

2) Select the appropriate study and click OK.

1) Click on the “factors filter” to open the filter window.

2) From the factor list select “GERMPLASM IDENTIFICATION – breeder name”.

3) Select the ‘in’ operator from the list.

4) Select the genotypes (mutants) you want to include in the figure: “M82”, “AB2”, “sft-4537”, “sft-4537 x M82”, “sft-7187”, “sft-7187 x M82”, “sft-stop”, “sft x M82”.

5) Click on “Add condition”.

Here I define the genotypes I want to include in the figure and add it to the conditions list.

1) Click on the “factors filter” to open the filter window.

2) From the factor list select “GERMPLASM IDENTIFICATION – breeder name”.

3) Select the ‘in’ operator from the list.

1) Click on the “factors filter” to open the filter window.

2) From the factor list select “GERMPLASM IDENTIFICATION – breeder name”.

4) Select the genotypes (mutants) you want to include in the figure: “M82”, “AB2”, “sft-4537”, “sft-4537 x M82”, “sft-7187”, “sft-7187 x M82”, “sft-stop”, “sft x M82”.

3) Select the ‘in’ operator from the list.

1) Click on the “factors filter” to open the filter window.

2) From the factor list select “GERMPLASM IDENTIFICATION – breeder name”.

Here I add another condition: use only genotypes that were not derived (segregated) from F2 family.

1) select “F2 segregation status”.

2) Select the ‘in’ operator from the list.

4) Click on “Add condition” to add it to the conditions list

3) Select the ‘0’ label which means that this genotype was not derived by segregation.

“Sft-stop x M82” plants in this study were segregated from F2 family. There were no plants of this genotype that were produced by a fixed cross of M82 and sft-stop. Nevertheless, we want to include them in our analysis. By adding another group of conditions we define the logic to be: condition1 AND condition2 in the first group OR condition3 from the second group. The logic operator is “AND” over all conditions within a group and “OR” among groups.

1) From the factor list select “GERMPLASM IDENTIFICATION – breeder name”.

2) Select the ‘in’ operator from the list.

3) Select ‘sft-stop x M82’.

4) Add another group of conditions

5) Click on “add condition” to the new group.

6) finally, click “OK” on the bottom of the window.

This is a group of conditions

After I’m done with the conditions, the “factors filter” icon is changed to filtered mode. Now I can select the “Fit Y by X” analysis (from the left) and put “Total yield” and “Brix” in the “Y variables” field and “GERMPLASM IDENTIFICATION – breeder name” in the “X variables”. Now I can decide to perform statistical comparison in order to compare all genotypes to the control – “M82”. For this I need to click on the “Compare means” link.

Compare means link

The “Compare means” opens a window for which I can define the control genotype and the statistical test in order to perform multiple means comparison.

1) Click on the compare means link

2) In the popup window select the desired test

3) Select alpha value

4) Select the control genotype.

5) Click “OK”.

Now I’m ready to invoke the analysis by clicking the ‘OK’ button.

Compare means parameters

1) Click ‘OK’ to perform the analysis

This is the output of the “Total yield” trait. The output displays the “M82” genotype in red bold which indicates that this is the level that all other genotypes are compared to. Then, “sft-stop” is in light-red, indicating that it is not significantly different than M82 under the selected a threshold - 0.05. all other genotypes are black, meaning they are significantly different than M82.

This is equivalent to figure 1B of Krieger et al.

This is equivalent to figure 1C of Krieger et al.

When you click the “Show table” button, it will display three tables (and then it is changed to “Hide table”).

Table1: shows all observation units. this is the raw data of the analysis

Table2: means and confidence interval of all genotypes

Table3: ANOVA