The current modal for displaying more information about the sequence optimization results appears cutoff. It needs to be completed and make a clear distinction between optimized_ad and optimized_sd.
Optimized SD v Optimized AD
They stand for "squared difference" and "absolute difference", respectively. From the code:
AD:
iterates upon the multi_table while optimizing the query to select the best-optimized DNA sequence using an
absolute-difference based method
SD:
iterates upon the multi_table while optimizing the query to select the best-optimized DNA sequence using a sum of
squares of differences based method
Basically, the difference is how the standard error is computed when adjusting the codon frequency table for multiple species. There are two ways that we can do this: 1) squared error, and 2) absolute error. Their respective formulas look something like this:
$$
\text{SD} = \left( y - \hat{y} \right)^2
$$
$$
\text{AD} = | y - \hat{y} |
$$
You can see that both produce positive values, but there seems to be debate about which is better. I didn't actually write this original code, however, so I am reluctant to make a definitive choice about which is better. Here is a stack exchange about which might be more appropriate. Notably:
The benefits of squaring include:
Squaring always gives a non-negative value, so the sum will always be zero or higher.
Squaring emphasizes larger differences, a feature that turns out to be both good and bad (think of the effect outliers have).
Therefore, it seems that squaring might be favorable and perhaps we can do away with the optimized_ad altogether. At least on the UI to make it clean and simple.
Problem
The current modal for displaying more information about the sequence optimization results appears cutoff. It needs to be completed and make a clear distinction between
optimized_adandoptimized_sd.Optimized SD v Optimized AD
They stand for "squared difference" and "absolute difference", respectively. From the code: AD:
SD:
Basically, the difference is how the standard error is computed when adjusting the codon frequency table for multiple species. There are two ways that we can do this: 1) squared error, and 2) absolute error. Their respective formulas look something like this:
$$ \text{SD} = \left( y - \hat{y} \right)^2 $$
$$ \text{AD} = | y - \hat{y} | $$
You can see that both produce positive values, but there seems to be debate about which is better. I didn't actually write this original code, however, so I am reluctant to make a definitive choice about which is better. Here is a stack exchange about which might be more appropriate. Notably:
Therefore, it seems that squaring might be favorable and perhaps we can do away with the
optimized_adaltogether. At least on the UI to make it clean and simple.