## page was renamed from Meetings/Nepal2016/Ara_GSM
= Practical 4 =
== Analysing a genome scale metabolic model of A. thaliana ==
In this practical we (you !) will be replicating some the analysis that was discussed in the previous lecture. In order to do this you will need to download the files associated with the model:

 1. Download the file [[http://mudsharkstatic.brookes.ac.uk/Nepal2016/P4/AraGSM.tgz|AraGSM.tgz]]  into the area in which you have been using for your other practicals.

 1. This is a compressed archive file and  you will need to extract the files before they can be used:
  .
  {{{ }}}{{{$ tar -zxf }}}{{{AraGSM}}}{{{.tgz }}}



 1. This will generate a directory, AraGSM, containing two  sub-directories: Model and Analysis. Model contains the model   definition files and an additional python module (in Model/Tools).      Analysis contains the python modules you will need for this     practical.

 1. For the sake of the practical we have made a few       simplifications and the model and results will not be identical to      those in the lecture. the aim of the practical is to illustrate the     techniques used.

== Part A: ==
== Investigating the effect of knocking out the Calvin cycle enzymes from GSM of A. thaliana ==
 1. cd into Analysis/Knockouts

 1.
 Start [[http://mudshark.brookes.ac.uk/ScrumPy|ScrumPy]]        and the load the model as before.


 1.
 Import the {{{KnockOut}}}      module.


 1.
 This defines a single function also    called {{{KnockOut}}}{{{Effects}}}      that returns a dictionary recording the impact of removing each   reaction from the model (relative change in objective value)


{{{#!python
   >>> import BuildLP
   >>> lp = BuildLP.BiomassLP(m)
   >>> lp.Solve()
   >>> wild_type_solution = lp.GetPrimSol()
}}}
Observe the fluxes of the Calvin Cycle enzymes

 1.
 Provide the model, reaction to be knocked out and {{{lp}} as argument to the  function {{{KnockOutEffects}}}


 1.
 Store the solution as {{{mutant_solution}}}


{{{#!python
       >>> mutant_solution  = KnockOut.KnockOutEffects(m, 'SEDOHEPTULOSE-BISPHOSPHATASE-RXN_Plas', lp)
}}}
 1. Investigate the differences in wild_type_solution and mutant_solution and interpret the results.
 1. Similarly , find the reactions that were inactive as a effect of knock out
 1. Identify the reactions that have different flux values as compared to wild type and interpret the results.

{{{#!python
      >>> from Util import Set
      >>> new_active_reactions = Set.Complement(mutant_solution, wild_type_solution)
      >>> inactive_reactions = Set.Complement(wild_type_solution, mutant_solution)
      >>> for rxn in Set.Intersect(mutant_solution, wild_type_solution):
      >>>      if abs(mutant_solution[rxn] - wild_type_solution[rxn]) > 1e-5:
      >>>            print rxn, mutant_solution[rxn], wild_type_solution[rxn]
}}}
 1.
 Repeat Step 5-8 for the {{{FBPase}}}, {{{PRK}}}, {{{G3Pdh}}} and {{{SBPas,  FBPase}}} dual knock out mutants.


== Part B :  Analysing the response of GSM of A. thaliana to varying input of photon flux ==
 1. Change directory to the relevant area:
  .
  {{{$ cd A.thaliana/Analysis/LightScan}}}



 1.
 Start ScrumPy and load the model:


{{{#!python

 >>> m = ScrumPy.Model("../../Model/AraTopLevel.spy")
}}}
(If you wish to avoid a bit of typing, leave the model name blank and use the file selector to find the model file instead.)

 1. Examine the files that are now presented - how much can you recognise from previous work in this course.

 1. Create an LP object with the GSM of A.thaliana as argument (see previous LP exercise for details on how to do this).

 1.
 Set the objective to minimisation of all fluxes. Since minimisation is the default direction, all you need to do is to use the {{{lp.SetObjective()}}} method with all reactions in the LP object as argument, which can be obtained as {{{lp.cnames.values()}}}


 1. Use the dictionary of biomass fluxes as fixed constraints.

{{{#!python
      >>> import BuildLP
      >>> lp = BuildLP.BuildLP(m)
      >>> fd = BuildLP.biomass_dictionary
      >>> lp.SetFixedFlux(fd)
}}}
 1. Try to solve the LP and make sure the number of reactions in the solution is non-zero

 1.
 You will scan the photon uptake reaction by fixing its flux to a set of  value in a linear range, specifically 50 points evenly distributed from 0  to 1. To do this, first generate a list of these values and store it as {{{collection}}} (Use the knowledge from exercise of the  [[http://mudshark.brookes.ac.uk/AccliPhot/WorkshopOne/prac1|python practical]]).


1. Import the module {{{DataSets}}} from {{{Data}}}. Create an instance of the {{{DataSet}}} class. The {{{DataSet}}} class is a subclass of the {{{ScrumPy}}} {{{matrix}}}  class, as are the stoichiometry matrices that you have looked at  before, so much of the structure and properties of data sets will be similar.

{{{#!python
      >>> ds = DataSets.DataSet()
}}}
 .
 1.Write a {{{for}}} loop over the list of values generated above. Inside the loop, set the constrain on the photon uptake reaction ({{{Photon_tx}}}) equal to the loop variable, solve the LP, collect the solution in a dictionary and update in a dataset. There is a {{{DataSet}}} method, called {{{UpdateFromDic(...)}}},  that updates the data set with a dictionary


{{{#!python
     >>> for number in collection:
     >>>      lp.SetFixedFlux({"Photon_tx" :number })
     >>>      lp.Solve()
     >>>      if lp.GetStatusMsg() == "Optimal":
     >>>           sol_dic = lp.GetPrimSol()
     >>>           ds.UpdateFromDic(sol_dic)
}}}
 1. You will now analyse the data. But before that, study the properties of dataset you have just updated. Please do not print the whole dataset.

1. Since we are interested in reactions that change flux as a response to  changes in photon flux, we need to identify these reaction in the data  set. We will do this by looking at the aboslute difference between the  smallest and largest flux through each reaction in the set. Use a {{{for}}} loop over the column names of the data set (if the data set is named as{{{ ds}}} this is {{{ds.cnames}}}). For each column name get the associated list of values (use the method {{{ds.GetCol(c)}}} with {{{c}}}  being the name of column, i.e. if you are in a loop, the loop  variable); calculate absolute of the difference between the maximum and  minimum flux value in the list (we can use the built-in functions {{{abs()}}}, {{{max()}}}, and {{{min()}}}  for this), if this difference is above a fixed threshold the reaction  (i.e. the loop variable) will be stored. Here is the code:

{{{#!python
         >>> changing = []            # list to store reactions with changing flux
         >>> lim = 1e-7               # flux threshold
         >>> for c in ds.cnames:
         >>>     col = ds.GetCol(c)
         >>>     if abs(max(col) - min(col)) > lim:
         >>>         changing.append(c)
}}}
 1.
 Use the plotting methods of {{{DataSet}}} to look at the flux responses of the changing reactions. Set the x-axis as the photon flux with the method {{{ds.SetPlotX(...)}}}, where the argument is a column name in the data set {{{ds}}}, here the photon transporter ({{{Photon_tx}}}) (note that {{{ds}}}{{{.SetPlotX(...)}}} only initialises the plot internally, you will not see the plot until data is added). Add columns to the plot using the method {{{ds}}}{{{.AddToPlot(...)}}},  where the argument is either a string (name of a column), or a list (of  column names). Similarly, you can remove column from the plot using the  methods {{{ds.RemoveFromPlot(...)}}} and {{{ds.RemoveAllFromPlot()}}}. You can check all the functions of the dataset  {{{dir(ds)}}}.


1. You can repeat the exercise using the {{{LightScan}}} module given in the {{{Analysis}}} directory.