Public RAW RNA-seq data Now What!!

[prussiap]

Hi guys,
This forum is great for a beginner, lots of how-to's for newbies and experts alike. I've seen snippets of what i want to do here and there but I was hoping to ask questions, engage the community and then at the end write a how-to/decision tree post for others to benefit from this post. If there is no interest let me know as well also sorry if it's in the wrong place. I'm new to this. Help me understand the process and others too .

Goal:
Take RAW RNA-SEQ data, understand quality control, trims if needed, discuss the different methods of alignment (or pipeline software), Discuss what's different between the processes of looking at splice variants, CNVs, exomes, expression profiling (maybe miRNA) at the different steps. I'm sure i'm missing things too.

I'll organize it all into a nice post later but let's get to it

So the idea is to start with RAW (spit out of the machine) data and from a public database and understand the data and what you can do with it. I decided on K562CellTotalFastqRep2_fastqc from ENCODE as it was suggested to me as a good GAIIX read. Description and Link.

To get the latter tier you need to select:
RNA-extract: Total RNA
View: FastqRd1,2
Platform: Illumina HiSeq 2000
Cell: HAoAF I chose fastqRd1
Rd1,2 stands for read 1,2 for bio replicates: I chose

I chose this data set because it's known, there is some experimental quality control, it's already been analyzed, it's large and unadulterated, and in theory total rna (though it seems there is nothing 200>)

Steps:
QC:
1. QC- I ran it in FASTQC and saw the attached file.

Trimming/Alignment:
1. Tophat/Bowtie/Cufflinks or using python/R?

Expression Profile: (Looking for Counts)
What tools and pathway?

Building Contigs/Exome: (looking for mRNA)
What tools and pathway?

Spliceosomes:
What tools and pathway?

At the end comparing with other samples


I leave many answers open (because of time and i'm also somewhat of a newbie ). This is more of an exercise so let's start with Alignment and understanding what you have. I'll edit this as we go. Help me make this more Useful and organized also.

[alexdobin]

ENCODE data

Hi prussiap,

you are talking about ENCODE data, not ENSEMBL, right?

The sample you have chosen is not a good example, it's one of the earliest samples we generated with an unusual library prep and sub-par sequencing quality. I would strongly recommend other samples such as whole cell poly-A+/- for K562 and other cell lines. ENCODE RNA-seq data was mapped with STAR: ftp://ftp2.cshl.edu/gingeraslab/trac...release/2.1.1/

If you have questions about ENCODE data, please send me a message.

Alex

[wupengpro]

Hi prussiap,

you are talking about ENCODE data, not ENSEMBL, right?

The sample you have chosen is not a good example, it's one of the earliest samples we generated with an unusual library prep and sub-par sequencing quality. I would strongly recommend other samples such as whole cell poly-A+/- for K562 and other cell lines. ENCODE RNA-seq data was mapped with STAR: ftp://ftp2.cshl.edu/gingeraslab/trac...release/2.1.1/

If you have questions about ENCODE data, please send me a message.

Alex

Hi Alex,

I have downloaded some ENCODE datasets from SRA in NCBI(http://www.ncbi.nlm.nih.gov/sra/SRX135162?&report=full). Are these ENCODE datasets raw data or clean data? Need I the further quality control? Which method of quality control do you recommend?

Thank you!

[alexdobin]

Hi Alex,

I have downloaded some ENCODE datasets from SRA in NCBI(http://www.ncbi.nlm.nih.gov/sra/SRX135162?&report=full). Are these ENCODE datasets raw data or clean data? Need I the further quality control? Which method of quality control do you recommend?

Thank you!

Hi @wupengpro,
the ENCODE data deposited in SRA is raw, filtered only by standard Illumina chastity filters. All of the data is clean and high quality, judged by high mapping rates (90-95%), high correlation of gene expression from bio-replicas (>0.98) and by correct clustering of the samples. I think you do not need any additional quality control or filtering of the .fastq files - however, it's always advisable to filter your alignments, for example, remove multi-mappers, non-concordant mates, non-canonical junctions.

[Richard Finney]

... it's always advisable to filter your alignments, for example, remove multi-mappers, non-concordant mates, non-canonical junctions.

This is interesting information to be throwing away.

[alexdobin]

... it's always advisable to filter your alignments, for example, remove multi-mappers, non-concordant mates, non-canonical junctions.

This is interesting information to be throwing away.

That is indeed interesting information for some applications, however, it also contains a significantly larger percentage of mis-mappings (i.e. false positives). I guess I need to re-formulate my statement more carefully: if the study does not involve (i) highly similar loci (e.g. paralogs), (ii) fusion/chimeric transcripts, or (iii) non-canonical splicing, it is advisable to remove (i) multi-mappers, (ii) non-concordant mates, (iii) non-canonical junctions.

[per_ngs]

Quality scores in fastqc for ENCODE RNASeq data

Hello,
I just downloaded the LHCN RNASeq data generated at Caltech. I merged the fastq files from the 3 runs to generate a single file and ran fastqc and I an a little bit confused about the output I have got. The per base quality graph in fastqc is showing quality score going upto 70 (attached) and the per sequence graph is showing peaks at approx 38 and 68 (also attached). According to the ENCODE documentation, the quality scores are phred 33, so how come the quality score graphs look like this?
Apologies if my question is silly and if i am not understanding the way fastqc works.

Thanks for help.
NGSnewbie

[Sujani]

hello all,

When I try to sequence 16s bacterial RNA using ABI 3130 it gives me heterozygous peaks. Since the microbes only contain a haploid set of chromosomes I am puzzled how it could be possible to indicate two peaks?
can someone please explain

[GenoMax]

Sujani,

You should create a new post for this question rather than this current thread. Perhaps one of the moderators can do it for you.

hello all,

When I try to sequence 16s bacterial RNA using ABI 3130 it gives me heterozygous peaks. Since the microbes only contain a haploid set of chromosomes I am puzzled how it could be possible to indicate two peaks?
can someone please explain

[Sujani]

GenoMax,

Im really sorry for the inconvenience.Unfortunately,Im finding it hard to post a new thread.

[GenoMax]

Once you log into SeqAnswers, click on the "Forum" link in the top left quadrant under "site navigation".

Select the appropriate forum to post in by clicking on the main title of the forum (e.g. core facilities).

On the page that opens next there should be a "new thread" button towards top left.

GenoMax,

Im really sorry for the inconvenience.Unfortunately,Im finding it hard to post a new thread.

[Sujani]

Once you log into SeqAnswers, click on the "Forum" link in the top left quadrant under "site navigation".

Select the appropriate forum to post in by clicking on the main title of the forum (e.g. core facilities).

On the page that opens next there should be a "new thread" button towards top left.

Genomax,

Thanks alot for the help!! I could post my issues as a new thread!!!

[cwzkevin]

Remember that you combined three runs results into one. It is very likely that the three runs does not have the same Phred offset. The mode at 38 is from Phred33 and the mode at 68 is from Phred64, obviously.
Now, you must combined the three runs this way:
Phred33 then Phred?? then Phred64
Such that when FastQC trying to guess the offset, all it can see are the codes from Phred33, and it concludes the data is from Phred33. FastQC doesn't use all reads in your data to guess, it only use 200,000 reads if I am correct.
After its conclusion of Phred33, FastQC keeps memorizing your data quality and maps the ascii codes based on Phred33. That is why the 2nd mode at 68 showing up.

http://en.wikipedia.org/wiki/FASTQ_format#Encoding

Hello,
I just downloaded the LHCN RNASeq data generated at Caltech. I merged the fastq files from the 3 runs to generate a single file and ran fastqc and I an a little bit confused about the output I have got. The per base quality graph in fastqc is showing quality score going upto 70 (attached) and the per sequence graph is showing peaks at approx 38 and 68 (also attached). According to the ENCODE documentation, the quality scores are phred 33, so how come the quality score graphs look like this?
Apologies if my question is silly and if i am not understanding the way fastqc works.

Thanks for help.
NGSnewbie

[per_ngs]

Hello Kevin,
Thanks for the response. I did find out from other sources on seqanswers that the data that i combined had data with different Phred offset. I ran fastqc on each of the files individually and noticed this as well. So, for now i am processing the data separately.
Regards,
NGSnewbie