DESeq Normalization and miRNAs Filtering
1
0
Entering edit mode
phmai1148 • 0
@a0d449cd
Last seen 54 minutes ago
United States

I have three questions:

  1. I want to use deseq2 only to normalize the miRNA-seq raw counts (I don't want to use it to do any kind of differential analysis/ downstream analysis). So, in order to conduct deseq data normalization (defined as counts divided by sample specific size factors determined by median ratio of gene counts relative to geometric mean per gene), shall I just use the following codes without using (normTransform) nor (vst) functions?

dds <- estimateSizeFactors(dds)

mat <- counts(dds, normalized=TRUE)

  1. My understanding is that the function (collapseReplicates) calculates the sum of the counts from technical replicates and put it into a single column of the count matrix. Is that correct? Should I perform this step before or after deseq normalization, or it does not matter?

  2. I also want to filter miRNAs based on a specific minimum number of counts in a minimum number of samples (e.g keep only those miRNAs with a minimum count of 10 in at least 50% of the samples). Should I do this filtering step before or after deseq normalization?

Thank you

DESeq2 Normalization • 166 views
ADD COMMENT
0
Entering edit mode

Thank you so much for your response.

And if I want to use deseq2 only to normalize the miRNA-seq raw counts (I don't want to use it to do any kind of differential analysis/ downstream analysis). So, in order to conduct deseq data normalization (defined as counts divided by sample specific size factors determined by median ratio of gene counts relative to geometric mean per gene), shall I just use the following codes without using (normTransform) nor (vst) functions? dds <- estimateSizeFactors(dds) mat <- counts(dds, normalized=TRUE)

ADD REPLY
0
Entering edit mode

normTransform gives you normalized data on log2 scale, vst gives variance-stabilized data on log2 scale. Use what you need for your purpose. Both run the estimation of size factors internally.

ADD REPLY
0
Entering edit mode

Thank you for the response. So, just to make sure I understood that correctly: -I will collapse the replicates using the (collapseReplicates) function before the deseq normalization using (counts, normalized=TRUE) or before the differential expression analysis using the (DESeq) function? -I will do the filtering step before deseq normalization using (counts, normalized=TRUE) or before the differential expression analysis using the (DESeq) function? -For the normalization process, do I have to use the (vst) function or just use (counts, normalized=TRUE) and (normTransform, f=log2, pc=1)?

I am asking these questions because I want to use DESeq2 only for normalization, not for the differential expression analysis.

ADD REPLY
0
Entering edit mode

The DESeq() function is described in the vignette, it also runs the normalization, so if you do the collapsing and prefiltering, and then either run the normalization manually and then get normalized counts, or run DESeq() and then get the counts -- it makes no difference. See vignette on getting normalized counts. Whether to use log2 normalized counts or the vst is on you, again see vignette, vst is generally a good choice for most downstream such as clustering etc.

ADD REPLY
0
Entering edit mode
ATpoint ★ 2.4k
@atpoint-13662
Last seen 1 hour ago
Germany

1) Yes, it's the sum, see ?collapseReplicates:

Collapses the columns in object by summing within levels of a grouping factor groupby

Do that before DESeq().

2) Also before DESeq(), see vignette on prefiltering.

ADD COMMENT

Login before adding your answer.

Traffic: 442 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