I am trying to calculate the Z-score for several columns with Log fold change (LFC) data from RNA-seq expression values. But the mean for all columns, except one of them, is -inf and the standard deviation is NaN. The reason might be "#NAME?" in different columns which was not available in the column that the mean and std values could be calculated correctly. I have 22269 gene and LFC values for them in each column.
Somehow, I found the answer but not completely. Some of expression values in treated cells are equal to zero. So, When I want to calculate the LFC, the values would be 0 or -1 (based on the two formula you choose for calculation of LFC ( first- (b-a)/a or second- (b/a) when b is treated and a is control)). Accordingly, the log2(0) and/or log2(-1) is not meaningful. So, it cause a "#NAME?" in excel file. Now the question is: Shall I replace all #NAME? with zero to calculate the Z-score?
Thank you in advance....
Hi Elyas,
To ignore infinitive data when you calculate log2, it is better to consider log2(data+1).
Soheil Yousefi Thank you very much Soheil. I have used (b-a/a) to calculate LFC. In this case the LFC values (for b=0) will be -1. So even log2(data+1) = log2(0) and the result= inf. What should I do in this case. Am I allowed to consider inf as zero in LFC list.
Soheil Yousefi I am not using RAW data for my project. I have retrieved the level 3 data from LINCS-L1000 project which is "Quantile-normalized gene expression profiles". Accordingly, I have calculated the log2(fold change) for control and treated. Afterward, I measured the Z-score for LFC data.
Also, I have considered following description of Fold Change for may analysis: "Fold change is a measure describing how much a quantity changes going from an initial to a final value. For example, an initial value of 30 and a final value of 60 corresponds to a fold change of 2 (or equivalently, a change to 2 times), or in common terms, a one-fold increase. Fold change is calculated simply as the ratio of the difference between final value and the initial value over the original value. Thus, if the initial value is A and final value is B, the fold change is (B - A)/A or equivalently B/A - 1. As another example, a change from 80 to 20 would be a fold change of -0.75, while a change from 20 to 80 would be a fold change of 3 (a change of 3 to 4 times the original)".
It might be wrong?
I guess if you use B/A, it should be correct.
I am not sure getting Z-score of log2 is correct, because log2 is a scaling of FC and z-score is a scaling method as well, so why did you apply scaling on FC data two times?
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python