results {DESeq2} parameters
1
0
Entering edit mode
Aurora ▴ 20
@aurora-15104
Last seen 5.4 years ago

Hi!

I am not sure I understand well the parameters alpha and lfc Treshold of this function : results {DESeq2}

I first understood that these parameters allows to choose a p value and log fold change threshold, for example if I chose a pvalue of 0.05 and a log fold change of two then I exptected to only keep the genes with a p-values >0.05 and an absolute log fold change >2.. 

I proceeded this way :

test_results=results(dds,contrast=c("Cellular_Type","Basal","LP"),lfcThreshold = 2, alpha=0.05)

Then I selected the genes passing the thresholds this way :

sresults_DE = test_results[which(test_results$padj <0.05 & abstest_results$log2FoldChange)>2),

Indeed, the function returns me all the genes without any selection so I don't really understand what are the purpose for this parameters

What will be the difference if I let them by default ?

 

 

deseq2 results • 1.1k views
ADD COMMENT
0
Entering edit mode
@mikelove
Last seen 12 hours ago
United States

Please check here first: ?results

We document details about every argument in these help files.

ADD COMMENT
0
Entering edit mode

i already read the details about these two arguments, but I didn't understood. For lfcThreshold is is writen :" a non-negative value which specifies a log2 fold change threshold." so I understood it as a threshold filter as well for alpha "the significance cutoff used for optimizing the independent filtering (by default 0.1)".

Sorry if it's a littble bit duv, I  would just like to know id it commanded to define the parameters if there are different from lfc 0 and alpha 0.01 for the DE genes threshold?

Thank you,

Have a good day

ADD REPLY
0
Entering edit mode

hi, 

A concise way to put it is that results() always gives you back a table with exactly the same number of rows as the dataset. 'alpha' sets the target FDR for the independent filtering, so this will only change the adjusted p-values. Filtered genes (due to low mean count) will get an NA in the padj column. 'lfcThreshold' changes the null hypothesis for the statistical test and so changes the p-values (which changes the adjusted p-values).

ADD REPLY
0
Entering edit mode

Ok thanks a lot ! So when you wannt only genes with a log FC between two conditions to be greate in absolute value than a threshold is it better to use the lfc Threshold parameter in the results function or is it ok if it's set to 0 by default ?

thanks a lot, 

have a good day

ADD REPLY
0
Entering edit mode

I recommend using the lfcThreshold argument, which will change the p-values, and then using an FDR filter. Now you have a gene set defined with an FDR bound, and the null hypothesis includes small fold changes, not just LFC=0.

ADD REPLY

Login before adding your answer.

Traffic: 806 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6