Which is the state-of-art, reliable and efficient method to obtain transcriptomics data, microarray or RNA-Seq?. Also can we get both mRNA and MiRNAs from one RNA-Seq experiment which is not possible with microarray as it needs 2 chips??. Does anyone have experience in analyzing RNA-Seq data, i have no clue. Though I saw many websites n tools which analyze microarray data. Any related papers, suggestions, ideas are greatly welcome. Thank you
RNA-Seq has the potential to outperform microarray, in theory at least. But practically it is not so clear cut. High statistical confidence results for differential gene expression is highly dependent on the number of biological replicates you run, regardless of which approach you take. Our lab typically uses 5 biological replicates per condition, but there are times when it would be nice to have even more. If you do not properly characterize the biological variation within each comparison group, you cannot derive robust statistical estimates of differential gene expression (there simply is no magical mathematical algorithm that can compensate for lack of biological replication to properly characterize the variance within each group in the contrast). DGE analysis in many ways is all about capturing and adequately accounting for the variance within each of your sample groups.
That problem alone often may push the cost of RNA-Seq well above performing the analogous experiment with microarrays. The other issue with RNA-Seq that remains unresolved is that of read depth - just how much sequence is necessary to meet or exceed the typical level and quality of results from a microarray experiment? That number is not at all clear (although it certainly seems to be considerable higher than earlier estimates published by the manufacturers of next-gen sequencers), and will likely vary a great deal depending on organism, tissue, experimental treatment and so on.
Lastly, there is no clear consensus at all on how to treat RNA-Seq data for differential gene expression analysis. In particular, there are numerous competing methods for normalization of RNA-Seq count data, and which is superior, or under what conditions one performs better than another, are not yet fully explored or even near so.
One place to keep an eye out for some answers to the analysis question will be the USA EPA's MAQC-III initiative (or the SEQC initiative as it is also called) which I believe is expected to publish its collective papers next year (I think they are planning to submit them as a group this fall?), see the web site:
http://www.fda.gov/ScienceResearch/BioinformaticsTools/MicroarrayQualityControlProject/default.htm
If you go with RNA-Seq, you really should investigate more than one normalization technique, and more than one differential gene expression statistical analysis (choices include conventional ANOVA, General Linear Models, Bayesian Statistics, Likelihood statistics, - you have numerous options). In terms of software, there are many options available, both open source (in R/BioConductor alone you have DESeq, edgeR, BaySeq, DEGseq and others) and commercial (e.g. Partek's new Flow package, JMP Genomics, and others).
We have some experience analyzing RNA-Seq data. RNA-seq is no doubt, better than Microarray on gene differential expression, mutation and gene structure analysis. one RNA-Seq experiment can provide all these information. There are many variants of RNA-Seq, such as poly(A)-positive mRNA-seq, poly(A)-negative RNA-Seq, small RNA-Seq. We used poly(A)-positive mRNA-seq to study the transcriptomes (here, we only focus on mRNA). As far as I know, miRNA and mRNA can not be obtained in a single RNA-Seq experiment, as they have different library preparation strategies. As for recommended papers, i don't have a good idea. However, my two publications are both related to mRNA-Seq analysis, and you can find them in my profile.
Thanks Jia-Yu for the reply. I will read your papers to get some ideas about the analysis part.
I'm glad I could help! Please don't hesitate to send message to me if you have any questions.
I agree that RNAseq is the way to go. For data analysis I suggest to use the Galaxy software package - freely available online @ https://main.g2.bx.psu.edu
RNA-Seq has the potential to outperform microarray, in theory at least. But practically it is not so clear cut. High statistical confidence results for differential gene expression is highly dependent on the number of biological replicates you run, regardless of which approach you take. Our lab typically uses 5 biological replicates per condition, but there are times when it would be nice to have even more. If you do not properly characterize the biological variation within each comparison group, you cannot derive robust statistical estimates of differential gene expression (there simply is no magical mathematical algorithm that can compensate for lack of biological replication to properly characterize the variance within each group in the contrast). DGE analysis in many ways is all about capturing and adequately accounting for the variance within each of your sample groups.
That problem alone often may push the cost of RNA-Seq well above performing the analogous experiment with microarrays. The other issue with RNA-Seq that remains unresolved is that of read depth - just how much sequence is necessary to meet or exceed the typical level and quality of results from a microarray experiment? That number is not at all clear (although it certainly seems to be considerable higher than earlier estimates published by the manufacturers of next-gen sequencers), and will likely vary a great deal depending on organism, tissue, experimental treatment and so on.
Lastly, there is no clear consensus at all on how to treat RNA-Seq data for differential gene expression analysis. In particular, there are numerous competing methods for normalization of RNA-Seq count data, and which is superior, or under what conditions one performs better than another, are not yet fully explored or even near so.
One place to keep an eye out for some answers to the analysis question will be the USA EPA's MAQC-III initiative (or the SEQC initiative as it is also called) which I believe is expected to publish its collective papers next year (I think they are planning to submit them as a group this fall?), see the web site:
http://www.fda.gov/ScienceResearch/BioinformaticsTools/MicroarrayQualityControlProject/default.htm
If you go with RNA-Seq, you really should investigate more than one normalization technique, and more than one differential gene expression statistical analysis (choices include conventional ANOVA, General Linear Models, Bayesian Statistics, Likelihood statistics, - you have numerous options). In terms of software, there are many options available, both open source (in R/BioConductor alone you have DESeq, edgeR, BaySeq, DEGseq and others) and commercial (e.g. Partek's new Flow package, JMP Genomics, and others).
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools