Metabolizer

User Tools

Site Tools

Trace:
in-silico_knockout:workflow_ko

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
in-silico_knockout:workflow_ko [2016/12/23 02:31] – ccubukin-silico_knockout:workflow_ko [2017/05/24 14:33] (current) – external edit 127.0.0.1
Line 8: Line 8:
 __**Using 1 sample:**__ __**Using 1 sample:**__
  
-**1)** Upload input data.+**1)** Upload //Input data//.
  
 <code> <code>
Line 18: Line 18:
 </code>  </code> 
  
-**2)** Select species. \\ +**2)** Select //Species//. \\ 
-**3)** Fill the job information.\\ +**3)** Fill the //Job information//.\\ 
-**4)** launch job.\\+**4)** Click the //Launch Job//  button\\
  
-Note: Uploading only 1 sapmle does not require design data and disables auto knockout.+{{ :in-silico_knockout:knockout2.png?600 |}} 
 + 
 +Uploading only 1 sapmle does not require design data and disables auto knockout. 
 +Fold change of the module activities between conditions will be shown as results of this analysis.
  
 __**Using dataset (multiple samples):**__ __**Using dataset (multiple samples):**__
Line 28: Line 31:
 Using multiple samples lets user to calculate class probability (before and after KO/OE ) of selected  sample. Probabilities are calculated using Random Forest machine learning algorithm. For unbiased learning  Using multiple samples lets user to calculate class probability (before and after KO/OE ) of selected  sample. Probabilities are calculated using Random Forest machine learning algorithm. For unbiased learning 
 sample sizes between classes might be balanced and sample size should be good enough. Once user upload design data, statistics of the design data used as criteria to let user do probability based analysis. sample sizes between classes might be balanced and sample size should be good enough. Once user upload design data, statistics of the design data used as criteria to let user do probability based analysis.
-This statistics and warnings appears after uploading design file.+This statistics and warnings appears after uploading design file. If the sample sizes are imbalanced the design file will be used only to select sample for KO/OE and only fold change of the module activities between conditions will be shown as results of this analysis.
  
  
-**1)** Upload input data.+**1)** Upload //Input data//.
 <code> <code>
 id sample1 sample2 sample3 id sample1 sample2 sample3
Line 39: Line 42:
 4 0.23 0.45 0.33 4 0.23 0.45 0.33
 </code> </code>
-**2)** Upload design data.+**2)** Upload //Design data//.
 <code> <code>
 sample1 Cancer sample1 Cancer
Line 45: Line 48:
 sample3 Healthy sample3 Healthy
 </code> </code>
 + {{ :in-silico_knockout:knockout.png?600 |}}
  
 if sample sizes are balanced and each group has more than 10 samples, you will get results of the class probabilities and also you will be able to use auto knockout option. if sample sizes are balanced and each group has more than 10 samples, you will get results of the class probabilities and also you will be able to use auto knockout option.
  
-**3)** Select species. +{{ :in-silico_knockout:knockout3.png?600 |}}
-**4)** If you want to knockout genes by one by and automatically, you can select this option. +
-For each gene you will get class probability including before and after KO. Even though this option executes KOs in parallel, the total analysis time can be up to 5 min.+
  
 +**3)** Select //Species//.\\
 +**4)** If you want to knockout genes by one by and automatically, you can select this option.
 +For each gene you will get class probability including before and after KO. Even though this option executes KOs in parallel, the total analysis time can be up to 5 min. \\
 +**5)** Fill the //Job Information// form and click the //Launch Job//  button.
in-silico_knockout/workflow_ko.1482460289.txt.gz · Last modified: 2017/05/24 14:33 (external edit)