FAQs
The app displays old data (OR)
Can’t uncheck the “load example” checkbox (OR)
Results from previous execution loads and executes automatically
These issues are related to the user’s web browser, due to a process called browser caching. One can solve this by using the following key combination while your browser is active.
For Chrome: Ctrl+F5 or Shift+F5
For Firefox: Ctrl+F5
For more information, check the following two links:
Is the normalization performed only for the selected columns in the table ?
Be it for the Analyse tab or Explore tab, the normalization is performed on the entire table where the values are integers (non-decimal data / count data). Therefore, it is highly recomended that one would remove those numeric columns from the data file that does not belong to the experiment especially if they are integers.
Good to Know: In order to avoid computational errors while dealing with the data containing zero values, a count of +1 is added by default to the entire data set (only integer columns) soon after importing the data. One can notice this from the table in the Import tab. If you are sure that your data doesn’t contain any zeros, or if you are only interested in using your data to produce plots in the Explore tab, you could opt-out of it by clicking the check box in the Import tab.
Exploring the Filtered Data.
As one has observed, the “Control Filter” in the Analyse Tab filters the resulting table only based on the controls (ctrl_a and ctrl_b). If you’d want to explore this filtered data set in the Explore Panel, you could download this result and reupload it and head to the Explore Panel, set the filter to zero and explore your data.
What does “calculate log2FC with mean counts” option do ?
Lets say, in the
Analysetab you have assignedctr_awith the samplest1.samp_r1andt1.samp_r2and you have assignedexp_awith the samplest2.samp_r1andt2.samp_r2where, \(t\) stands fortime pointand \(r\) stands forreplicate.
1. calculation “with means”
In the checked state (figure above): the log2 fold changes between ctrl_a and exp_a will be calculated with the mean counts of the samples assigned to them.
\[ctrl\_a = \frac{t1.samp\_r1 + t1.samp\_r2}{2}\] \[exp\_a = \frac{t2.samp\_r1 + t2.samp\_r2}{2}\] \[\mathbf{par\_a} = log2\Big(\frac{exp\_a}{ctrl\_a}\Big)\]
These simplified equations explain the calculation of par_a in the case of the above example. When the slider limits are set for \(\mathbf{par\_a}\) , with means mode picks those candidates that satisfy these limits for \(\mathbf{par\_a}\). The other parameters par_b, par_d and par_e are calculated similarly. The par_c is calculated as,
\[par\_c = par\_a - par\_b\]
2. calculation “per replicate”
\[par\_a_1 = log2\Big(\frac{t2.samp\_r1}{t1.samp\_r1}\Big) \] \[par\_a_2 = log2\Big(\frac{t2.samp\_r2}{t1.samp\_r2}\Big) \]
When the slider limits are set for \(\mathbf{par\_a}\), per replicate mode ensures that it picks only those candidates that satisfy these limits for both \(par\_a_1\) and \(par\_a_2\). The other parameters par_b, par_d and par_e are calculated similarly. The par_c is then calculated as,
\[par\_c_1 = par\_a_1 - par\_b_1\] \[par\_c_2 = par\_a_2 - par\_b_2\]
NOTE: One will notice that even when the program runs in per replicate mode, the par_a, par_b …etc. values in the result table contains the values as if they were run by with means mode. This is done purely for the sake of convinience because, if there are n number of replicates, the program would generate n number of parameter values and thereby the table would expand so much that it makes no sense to look at those values. Therefore, the result table has the abstract values although the behaviour of the slider limits is different for the selected modes.