deletions | additions       diff --git a/To_have_a_fair.tex b/To_have_a_fair.tex  index 84a0613..d8b3186 100644  --- a/To_have_a_fair.tex  +++ b/To_have_a_fair.tex 
...
\begin{table}  \begin{tabular}{ c c | c | p{3cm}p{8cm}}  \hline  Seg&Replicate (E)&$d_0$&Signal $S_3^{d_0}$&Noise $S_3^{d_0}replace_contentNoise  $S_2^{d_0}$\\ \hline  1 & 1 & 8.713 & $\emptyset$ & 33, 281, 404, 824, 839, 926, 1889, 2290, 2298, \textcolor{blue}{2299}, \textcolor{blue}{2303}\\  & 2 & 9.138 & 2072 & 311, \textcolor{blue}{2299}, \textcolor{blue}{2303}\\ 
...
Taking the intersection of the findings from both replicates, we identify only S6-822 as a substitution site due to the treatment, along with S1-2299, S2-2303, S3-173, S3-174, S3-176, S3-176, S3-200, S3-203, S3-2078, S3-2192, S3-2193, S3-2195, S5-24, S5-389, S5-1103, S6-977 as the locations with large variation not due to the treatment.   Comparing the result in Foll et all al  \citep{Foll2014} to our findings (Table \ref{tab:H1N14t}), we notice that all the extra locations identified there appear in the result for the first biological replicate, but not in the result for the second. This phenomena is likely caused by the fact that the average read count per site for the second replicate is much less smaller  than the first. The analysis in Foll \citep{Foll2014} based on a  population geneticbase  approachis  therefore biased appears to be heavily influenced  by the first replicate. It This  leads us to postulate that their result is adversely affected by the large imbalance in counts. With addition samples, we Noticing that the two additional sites identified in Table \ref{tab:H1N15t} show more pronounced drug effect after the 12th passage (see Figures \ref{fig:S5-300},\ref{fig:S8-80}). We  conclude that IVA had not fully responded to the treatment by Passage 12 and our previous analysis, including the last time point collection, is more reliable for the identification of substitution sites.