kundajelab / DMSO

5 stars 3 forks source link

RNA-seq voom mean-variance trend #1

Open annashcherbina opened 7 years ago

annashcherbina commented 7 years ago

Voom(counts) converts raw counts to log-cpm values:

voom_mean_variance_trend

Raw counts + surrogate variables added to model

1) limma_voom(counts) -> Model + SVASEQ_vars

sva_adjusted_model_lmfit_with_sva_contribs

2) limma_voom(counts) -> Model + exp(SVASEQ_vars) sva_adjusted_model_lmfit_exp_sva_contribs

asinh(tpm) + surrogate variables added to model

3) limma(asinh(tpm))->Model+SVA_vars

sva_adjusted_model_lmfit_sva_contribs_starting_with_tpm

4) limma(asinh(tpm))-> Model+exp(SVA_vars) sva_adjusted_model_lmfit_exp_sva_contribs_starting_with_tpm

asinh(tpm) - surrogate variables

5) limma(asinh(tpm)- sur vars) sva_subtracted_from_input_tpm 6) limma(asinh(tpm) - exp(sur vars)) exp_sva_subtracted_from_input

rlog(counts) --> Model + surrogate variables

7) rlog_sva_subtracted_from_input

8)

exp_sva_subtracted_from_input_rlog 9)

10 exp_sva_subtracted_from_input_rlog )

rlog(counts) - surrogate variables