search

kundajelab / DMSO

5 stars 3 forks source link

rlogTransform(counts) without sva versus asinh(tpm)-SVA #5

Open annashcherbina opened 7 years ago

annashcherbina commented 7 years ago

not performing SVA on the rlogTransform(counts) data does improve concordance with the asinh(tpm)-sva model. I also lowered the FDR threshold to 0.1, and the LFC threshold to 0.5 for the rlogTransform(counts) model:

Volcano plots

volcano earlyg1 volcano earlyg1_vs_gs2m_c volcano earlyg1_vs_sg2m_t volcano lateg1 volcano lateg1_vs_earlyg1_c volcano lateg1_vs_earlyg1_t volcano sg2m volcano sg2m_vs_lateg1_c volcano sg2m_vs_lateg1_t

Venn diagrams

venn earlyg1 venn earlyg1_vs_gs2m_c venn earlyg1_vs_sg2m_t venn lateg1 venn lateg1_vs_earlyg1_c venn lateg1_vs_earlyg1_t venn sg2m venn sg2m_vs_lateg1_c venn sg2m_vs_lateg1_t