how to choose a better tight value '-t' ?

[Ruismart]

hi, glad to know this nice tool !
I wonder

1. if there's anyway to decide which '-t' value better
2. which input is better, all expressed genes (like GSEA) ? or merged filtered DEGs (p<0.05, fold change > 2) ?

I'm exploring a RNAseq dataset with 3 time points, and I have ranged '-t' from 1 to 10, the pattern change a lot, even a similar pattern in different '-t' has different number of genes in it. should I just select one only by visualization ?

the two kinds of input would also affect patterns and number of genes in each pattern.