Trouble writing Pulp constraint

Question

I am trying to solve a LP problem, I have a dataframe of 100 rows and 2 columns( let's say a, b) and I need to choose 10 rows such that following conditions should meet:

if a>=b then b/(a+b)>=0.2
if a<=b then a/(a+b)>=0.2

I tried following steps( Suppose dataframe is df)

1. Made a dummy variable c which contains only value 1 in each row and key representing each unique row.
2. Then key=list(df['key'])
3. a=dict(zip(key,df['a']))
4. b=dict(zip(key,df['b']))
5. c=dict(zip(key,df['c']))
6. var=LpVariable.dicts('keys',key,lowBound=0,cat='Binary')
7. prob= *defined minimizing function*
8. prob+=lpSum([c[i]*var[i] for i in key)<=10
9. prob+=condition(lpSum([a[i]*var[i] for i in key),lpSum([b[i]*var[i] for i in key))=1

def condition(a,b):
   if a>=b and b/(a+b)>=0.2: return 1
   if a<=b and a/(a+b)>=0.2: return 1
   return 0

I am having trouble with the last step. Is there any work around?

Note: I understand there can be some other methods or tools to solve this kind of problems, but I have a restriction to go with Pulp since this is just a part of a bigger problem, so any guidance with this will be helpful.