I would like to know how to decide the minFraction parameter in the preprocessing step (xsms groupChrom Peaks group), because it is indicated that we have to choose it depending on the class column but in this column I only have three categories (pool; blank and sample) and there is no distinction between samples (it is in the condition column). For example, if I have 16 conditions and 8 replicates by condition, which should be the correct parameter ?
You can add a column called 'class' in column 2 of your sampleMetadata file with for each sample on of your 16 conditions named as for example cond1, cond2..., you will also have QC and blank, the 3rd column will be your sampleType with pool, blank and sample categories. This will allows minFrac to take into account your design. Then depending on your needs you xan set minFrac to 0.5 meaning that you want to keep features detected in 50 % of the samples of at least one class. Or you can increase the value if you want to be more stringent. If you set minFrac to a low value like 0.01 then you will keep a lot of features detected only in some samples. This is for example sometimes usefull for phenotyping experience where people search for unique compounds produced by one strain over many. But in your case I think that you want consistent signal inside a condition, so I would go for 0.8 or even 1 if you want to be really strict
Hope it helps
1 « J'aime »