Problems with QC results of Illumina Body Map 2.0

[hajime]

Dear all:

I'm new in the sequencing analysis.

Recently, I've downloaded the RNA-seq data of Illumina Body Map 2.0.
(The accession number is GSE30611.)

The data was applied this following processes:
(1) All the SRA files were converted to Sanger FASTAQ.
(2) Sanger FASTQ files without any manipulation was uploaded to Galaxy.
(3) Fastx toolkit was used to draw the quality plot of these sequences.

Finally, I got some strange results of quality control.
For example, both of single-end and pair-end sequence from Brian tissue shows bad QC from the first base to 10th base nucleotide (shown in following figures). It's not a typical QC plot for Illumian sequencing.

Single-end:

Pair-end:


I think there may be some problems in library-constructing step.
Does anyone can give me any idea why I got these results?

Thanks!

Best,
Yi

[kopi-o]

I believe the pattern of lower quality scores in the beginning is quite common in Illumina data. Overall, these quality scores are remarkably good. The median never drops below 30!

[hajime]

I believe the pattern of lower quality scores in the beginning is quite common in Illumina data. Overall, these quality scores are remarkably good. The median never drops below 30!

Thanks for replying!

In facts, I've never seen this kind of data before...
The data I processed before contained lower quality at the end of sequences only...

Therefore, I feel confused...

Why or how does the Illumina generate the lower quality score in the beginning sequences?

Thanks again!

[kopi-o]

See this thread:

http://seqanswers.com/forums/showthread.php?t=13155

[hajime]

Excuse me, there is another question here:

The first quantile (25%) for the first 10 bases in this data is lower than Q30 even Q20. Indeed, that means there are 25% reads contains bad q-score in the first 10 bases, doesn't it?

Should I trim the first 10 bases that with lower quality out for all sequences at the beginning of data processing? or trim out only the sequences that contain lower q-score?