Hi all,
I have a fundamental question that is a bit embarrassingly rudimentry, considering I have been investigating microRNAs for a couple of years now.
I'm working with data from the TCGA and I am hoping to somehow gain insight about how reflective normalized values (e.g., transcripts per million mapped microRNAs) are of the true microRNA levels among different tumors. Is there any way to normalize across tumor samples to better compare these normalized microRNA levels? It seems like people generally assume that this can be done in many cases, but it is not at all clear to me how valid the assumption is. I know little about how sample preparation and degradation come into play to determine final levels.
Follow-up: Given that with TCGA data sets, in many cases the microRNAseq preparation and RNAseq preparation are carried out using the same aliquot, and given that RNAseq data can be more easily normalized across tumors using stably-expressed genes (or groups thereof, as with Cibersort), would some normalization method for the microRNAseq data then be possible based on RNAseq levels for these stable genes from the same aliquot? It seems like this would at least normalize some aspects of the bias.
Thanks in advance for any thoughts on this subject.
Matt
I have a fundamental question that is a bit embarrassingly rudimentry, considering I have been investigating microRNAs for a couple of years now.
I'm working with data from the TCGA and I am hoping to somehow gain insight about how reflective normalized values (e.g., transcripts per million mapped microRNAs) are of the true microRNA levels among different tumors. Is there any way to normalize across tumor samples to better compare these normalized microRNA levels? It seems like people generally assume that this can be done in many cases, but it is not at all clear to me how valid the assumption is. I know little about how sample preparation and degradation come into play to determine final levels.
Follow-up: Given that with TCGA data sets, in many cases the microRNAseq preparation and RNAseq preparation are carried out using the same aliquot, and given that RNAseq data can be more easily normalized across tumors using stably-expressed genes (or groups thereof, as with Cibersort), would some normalization method for the microRNAseq data then be possible based on RNAseq levels for these stable genes from the same aliquot? It seems like this would at least normalize some aspects of the bias.
Thanks in advance for any thoughts on this subject.
Matt