According to the sample data you posted:

0.95 0.02 0.07 0.07 #A
0.03 0.01 0.06 0.83 #C
0.01 0.02 0.80 0.09 #G
0.01 0.95 0.07 0.01 #T
[download]

Each column adds up to 1.00, but each row adds up to an arbitrary value. Since you're putting 0.95, 0.02, 0.07 and 0.07 into the A array, it makes sense that the A array adds up to 1.11 :)

PS: Why are you adding 0.0001 to your data? Printing the output with appropriate printfs should round the values off nicely and hide the artifacts from the floating point math in the CPU.