Trimmomatic: Quality Trimming read loss -

[Patrice]

Hello community,

I plan to do a reference assisted genome assembly for a yeast genome.
I have some questions regarding the trimming prior the assembly:

# Setup:
Illumnia MiSeq / Paired end sequencing / coverage around 80x

# Case:
Intended pipeline:
raw reads --> trimming --> Spades or Velvet --> Contiguator

I ran FastQC on the raw reads (see attached files, forward reads shown):
FastQC does not detect adapters.
Since the quality drops in the end of the reads, so I decided to use trimmomatic for quality trimming.

I needed to apply a strict quality cutoff in order to obtain better FastQC results (see attachements). I used following parameters in trimmomatic:

SLIDINGWINDOW:4:30 HEADCROP:11 MINLEN:40

However, I recognized a huge loss of reads after the trimming :

Raw reads: 9322401
trimmed reads: 7320129
-> loss of 21 %

# Questions:
1. Since FastQC did not detect illumina adapters in the raw reads, would you still run a adapter removing step (eg AdapterRemoval or trimmomatic [ILLUMINACLIP])?

2. What would you suggest regarding the quality trimming? Some people suggest not to run any trimming at all. Should I go down with the cutoff values in order to keep more reads?

Thank you and best regards

[Patrice]

I calculated the Coverage:

After trimming
Coverage = (7320129 reads * 130 mean reads)/ 20.500000 bp = 46,8x

So I suppose that is still fine even after trimming

[GenoMax]

It would be wise to scan (and trim as needed) the data for presence of extraneous sequences. Since you are going to de novo assembly, you don't want any other sequence that does not belong to your genome in there.

You could trim at Q20 to begin with and see what you get. If you feel you have plenty of coverage (which you seem to think) then you could go with the ultra-stringent Q30 trim.

[Gorgon_VZ]

Having a look on your FastQC figures I recognised, that even before your trimming approach you have a huge number of short reads around 90bp (I guess the first figure shows the data before trimming and the second one afterwards!?)! Are you sure that your data is rawdata? Regarding the adapters you should also check if your adapter is present in the FastQC "adapter_list.txt" file. If this is not the case, you can provide a custom file via --adapters option. Looking at the kmers is also a good indicator of possible artificial sequence contamination!