about RNA-Seq data from cell lines

[alittleboy]

I am currently analyzing some RNA-Seq data from different cell lines, but I think I lack basic knowledge in this area. Are there any related papers/online tutorials on this subject? For example, I'd like to know:

(1) if different cell lines are from different subjects;
(2) how to deal with technical replicates of a cell line;
(3) under what situations can I compare different cell lines to see if there is, say, differential exon/isoform usage;
(4) ... and questions like that...

Thanks a lot!!

[GenoMax]

You do not clearly say if you are just beginning with RNA-seq data analysis or need help with just those specific questions. In any case perhaps this could help some: http://en.wikibooks.org/wiki/Next_Ge..._%28NGS%29/RNA

http://www.nature.com/nprot/journal/v7/n3/full/nprot.2012.016.html

[alittleboy]

You do not clearly say if you are just beginning with RNA-seq data analysis or need help with just those specific questions. In any case perhaps this could help some: http://en.wikibooks.org/wiki/Next_Ge..._%28NGS%29/RNA
http://www.nature.com/nprot/journal/....2012.016.html

Thank you for the resources! I think I've done some statistical analyses on RNA-Seq data, but lack the experience of real-world designs as I have the questions listed above. Maybe some more information on RNA-Seq analysis from cell lines are greatly apprecipated ;-)

[dietmar13]

i really don't understand your questions? what do you want know from your cell lines? you can't start an analysis without a biological/scientific question...

1) cell lines are from different individuals and different tissues (normal/cancer, the former mostly immortalized). some are genetically or with genetical selection modified, all harbor some cell culture artefacts (losses and gains of genetic material, changes in epigenetics, ...)
2) technical replicates are more or less useless, you would need biological replicates. are you sure, that your samples are tech. repl.
3) why do you want to know this - for sure are there differences, especially if the cell lines are from different tissues/conditions.
4) tell us some names of your cell lines, than we could try to understand the data set.

but as said above, EVERY analysis should have a clear biological question...

[shi]

Hi @alittleboy,

Not sure if this helps, but here is a case study for comparing RNA-seq data generated from two cell lines (UHR and HBR):

http://bioinf.wehi.edu.au/RNAseqCaseStudy/

Best wishes,

Wei

[alittleboy]

i really don't understand your questions? what do you want know from your cell lines? you can't start an analysis without a biological/scientific question...

1) cell lines are from different individuals and different tissues (normal/cancer, the former mostly immortalized). some are genetically or with genetical selection modified, all harbor some cell culture artefacts (losses and gains of genetic material, changes in epigenetics, ...)
2) technical replicates are more or less useless, you would need biological replicates. are you sure, that your samples are tech. repl.
3) why do you want to know this - for sure are there differences, especially if the cell lines are from different tissues/conditions.
4) tell us some names of your cell lines, than we could try to understand the data set.

but as said above, EVERY analysis should have a clear biological question...

Thank you so much for the reply! Yes, I agree that biological question comes first before discussing these details. We would like to compare different approaches for testing differential exon usage, and the data we have are many RNA-Seq cell line outputs, from which some of them have replicates (technical reps). I will talk to the bioinformatics people later for clarifications.

[alittleboy]

Hi @alittleboy,

Not sure if this helps, but here is a case study for comparing RNA-seq data generated from two cell lines (UHR and HBR):

http://bioinf.wehi.edu.au/RNAseqCaseStudy/

Best wishes,

Wei

Hi Wei:

Thanks for your reply! I read through the tutorial you wrote, where "A_1, A_2, B_1 and B_2. A_1 and A_2 are both Universal Human Reference RNA (UHRR) samples but they underwent separate sample preparation." -- do you mean that A_1 and A_2 are biological reps (and same for B_1 and B_2)?

If I understand correctly, cell type A is related to one tissue (say brain) with cancer but _1 and _2 are from two subjects?... how about cell type B? same/different tissue? same/different cancer?

Thanks again ;-)

[shi]

Hi @alittleoboy,

A_1 and A_2 are not biological reps in the strict sense. We refer as biological replicates those samples taken from different mice or taken from different pools of mice. A_1 and A_2 underwent different sample prep processes and were then sequenced. The sample variations between them should be higher than those typically seen in the technical replicates (undergoing same sample prep process).

Both samples A and B are from commercial cell lines. Sample A is Universal Human Reference RNA including RNAs extracted from 10 different types of cancer cells. Sample B is a human brain sample.

See the links below for more info about these samples:

http://www.genomics.agilent.com/en/References-Controls/Universal-Reference-RNAs/?cid=AG-PT-172&tabId=AG-PR-1217

http://products.invitrogen.com/ivgn/product/AM6050

Best wishes,

Wei

[alittleboy]

Hi @alittleoboy,

A_1 and A_2 are not biological reps in the strict sense. We refer as biological replicates those samples taken from different mice or taken from different pools of mice. A_1 and A_2 underwent different sample prep processes and were then sequenced. The sample variations between them should be higher than those typically seen in the technical replicates (undergoing same sample prep process).

Both samples A and B are from commercial cell lines. Sample A is Universal Human Reference RNA including RNAs extracted from 10 different types of cancer cells. Sample B is a human brain sample.

See the links below for more info about these samples:

http://www.genomics.agilent.com/en/References-Controls/Universal-Reference-RNAs/?cid=AG-PT-172&tabId=AG-PR-1217

http://products.invitrogen.com/ivgn/product/AM6050

Best wishes,

Wei

Hi Wei:

Thanks for the information! I agree with you that they are not biological replicates in the strict sense. It's better to call them technical replicates, even though they have higher variations than replicates having the same preparation procedures.

Here is the question: does it make sense to make such comparisons between non-biological replicates? I once heard that technical replicates are more or less useless, and biological replicates are crucial. Do we proceed with our analyses by whether it's a comparison between biological replicates, or by the variations relative to true technical replicates (in other words, if there is higher variation, we don't care if they're biological reps or not -- it's valid to compare)?

Thanks again ;-)

[shi]

Hi @alittleboy,

Yes, you should have biological replicates included in your experiment. The biological replicates are important for estimating the biological variations in your data and for confidently calling differentially expressed genes.

This case study just showcases the use of an easy-to-use and powerful Bioconductor pipeline for analyzing RNA-seq data. Although there are no biological replicates included in the case study, the pipeline remains the same for the analysis of RNA-seq data with biological replicates.

Best wishes,

Wei

[alittleboy]

Hi @alittleboy,

Yes, you should have biological replicates included in your experiment. The biological replicates are important for estimating the biological variations in your data and for confidently calling differentially expressed genes.

This case study just showcases the use of an easy-to-use and powerful Bioconductor pipeline for analyzing RNA-seq data. Although there are no biological replicates included in the case study, the pipeline remains the same for the analysis of RNA-seq data with biological replicates.

Best wishes,

Wei

Hi Wei:

Thank you for the clarifications! You're right that this is only an illustration. In reality, biological replicates are crucial (which I realize that biologists do not realize the importance, that's why I asked if comparing just tech reps are valid...)