What is the Reference to be used for RNA-Seq assembly?

[rohitngs]

Hello all.

I have been trying to do a Reference-based assembly for my RNA-seq data. It is Illumina paired-end data of 100bp reads.
There are 8 different sets of data samples.

But now I have a problem for aligned the reads to the reference. I have the fasta sequence of contigs, scaffolgs, genes and complete mRNA. The sequences are split in the form of chromosomes. This data was downloaded from an already available browser, with view in Chromosome-wise.

I plan to use Bowtie for the alignment, and I need a new Index for my genome as its not readily available.
Which is the reference for Transcriptome alignment - contigs, scaffolgs, genes, complete mRNA or something else?

I tried using contigs, mRNA, scaffold fasta sequences but only 0.05% of the reads align to any of these.
What exactly must be used as the reference? What should the Bowtie Index be built upon???

My aim is to create the BAM files so that it can be viewed in the Genome Browser.

[NicoBxl]

You have to use the genome as reference in a reference-base assembly.

[rohitngs]

The complete genome is not available; only the contigs, scaffolgs, genes and complete mRNA are available

[rohitngs]

I am trying to use the scaffolds that are currently available, is that fine???
Will that be ok to align my RNA-seq reads???

[NicoBxl]

Maybe you can try De-Novo assembly

[rohitngs]

The problem with de-novo assembly was I could not generate the SAM files properly...

[NicoBxl]

In general a De-Novo assembly will output a list of transcript in fasta file

[rohitngs]

Yes, I tried Rnnotator and got the final contigs.
But I could not generate the BAM files from that as it always showed no header line error

[NicoBxl]

Maybe try Trinity, it worked well. On which organism do you work ?

[rohitngs]

Silkworm, it has very high repetitive sequences... and the sex chromosomes have not yet been sequenced...

[mbayer]

Hi rohitngs,

if your genomic assembly is fairly complete you could try to map the reads onto that (combine the contigs and scaffolds into a single FASTA file and index that for Bowtie).

In theory you could also map your RNASeq reads to, say, a set of full length cDNA sequences, but you would need to be confident that it is pretty complete (and you never really can be), or you will throw away a lot of data.

However, I am alarmed by the fact that only 0.05% of your reads mapped when you mapped them to the genomic assembly. That to me suggests that your data is not from silkworm at all, and I would first check that it hasn't been contaminated with something else or even swapped with another sample. We had a case like this here not so long ago.

Just take a few reads and BLAST them against the NCBI database - you will quickly get an idea of what is going on.

cheers

Micha

[rohitngs]

Yes, that was the reason...

I was working with miRNA pooled libraries thinking that they were actually mRNA sequences... I did not know of that miRNA library data existed till I asked them if there was any other RNA-seq done...
I got it last week only when I was looking through all the data I was given, forgot to update that...

It was pretty stupid of me, but I guess that's how I had to learn...
Thank you so much Micha....

Cheers,
Rohit