Extent of reduction during pre-processing

[cerebralrust]

I'm just curious as how much reduction others observe during pre-processing of ngs data.

I've got 454 rna-seq data of about 300,000 reads.
Based on the fastqc report and using fastx toolkit,
- removing reads with bases below quality score of 20
- removing reads containing Ns
-removing ribosomal rna sequences (167 using bwa)
-removing reads below 100bp length

the dataset reduced to 185,735. I felt like this is too small. Is such a reduction common?
I could retain the reads less than 100bp in length , align them separately and align the entire dataset again.

I appreciate any advice and/or insight. Thank you.

[westerman]

It does seem like you are being rather strict. While it depends on your project and the program you wish to use, most NGS programs are rather insensitive to poor quality and short reads. Thus I rarely eliminate these. Of course rRNA can mess up a project as can adapter sequences so it can be useful to remove (or trim) reads that match those sequences.

[cerebralrust]

Thank you very much for your reply, Rick!

I'm using both MIRA and Trinity for assembly. Trinity is supposedly less sensitive to low quality compared to MIRA.

Anyhow, i will stick to removal of rRna and trimming adaptor sequences prior to assembly.

Thanks a lot!

[Manal]

Hi All
I have one basic question, I am new in the NGS field / Illumina and little bit confused about the meaning of Phasing and Prephasing rate and how does it happen?
Can anyone explain ?

Thanks

[mastal]

See the explanation in the Rationale section of this paper:

http://genomebiology.com/2009/10/8/R83

Basically phasing/prephasing refers to the problem that not all the molecules in a cluster on the flow cell will successfully incorporate a base during each cycle, or some may incorporate more than 1 base.