Question: Cross platform normalization for miRNA data
0
gravatar for Prasad
4.0 years ago by
Prasad1.6k
India
Prasad1.6k wrote:

Hi all,

I have to compare 2 miRNA data of our lab (NextSeq500, 75bp, ~10M reads/sample) to public data which is sequenced using GAIIx platform(data). I am performing sample clustering based on the miRNA expression (normalized data using DESeq) and not getting desired clusters. Are there any R packages or tools for cross platform normalization before comparing the data. Found some tool like AnyExpress, ComBat, not sure for my case

Any suggestions on tools/ way to compare the data

thanks

mirna tool R • 995 views
ADD COMMENTlink modified 4.0 years ago • written 4.0 years ago by Prasad1.6k

ComBat will remove clustering by technology ... that is what it is designed to do, but in doing so it will also remove and clustering by any experimental factor that correlates perfectly with technology. You want to be very careful with miRNA results as there is generally accepted to be quite a lot of sequence specific bias (i.e. GC bias etc) in miRNA results, and I'd get nervous about something with large sequence bias and different technologies (where the biases are likely to be different).

ADD REPLYlink written 4.0 years ago by i.sudbery9.7k

GC bias is an issue in case of miRNA. Any suggestion for my case to normalize the data? Is it really necessary to normalized data from two different platform?

ADD REPLYlink written 4.0 years ago by Prasad1.6k

" Is it really necessary to normalized data from two different platform". Well given that you are not getting the clustering expected, I'd say yes.

ADD REPLYlink written 4.0 years ago by i.sudbery9.7k

thank you. any tool or a R package to normalize?

ADD REPLYlink written 3.9 years ago by Prasad1.6k
0
gravatar for YaGalbi
4.0 years ago by
YaGalbi1.5k
Biocomputing, MRC Harwell Institute, Oxford, UK
YaGalbi1.5k wrote:

Not sure if this applies here...but I have been measuring tissue specificity across different tissues. While the platforms used to process each tissue is the same, the principal is that I had to normalise RPKM values across different experiments (in your case, across different platforms).

I did this by quantile normalizing the data using the 'normalize.quantiles' function of the preprocessCore library in R

Maybe this will lead you in the right direction.

ADD COMMENTlink written 4.0 years ago by YaGalbi1.5k

i have tried quantile normalization and row wise median normalization as mentioned here, not worked

ADD REPLYlink written 4.0 years ago by Prasad1.6k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1282 users visited in the last hour