what percentage range is normal for undetermined reads in a lane?

[zeam]

Hi,

In the demultiplexed results, there are some undetermined reads. I saw there is one lane that 53% 'of the raw clusters per lane' were undetermined. I'm surprised of this high percentage.

I'm wondering what percentage range is normal for undetermined reads.
Is there any big different for this for different sequence libraries (for example, BS-seq, resequencing, RNA-seq, etc)? Thanks in advance.

Shaojun

[nucacidhunter]

It depends on at least three factors:
1- PhiX% spike-in as the library does not have an index
2- quality of library adapters which will be different depending on supplier. Any miss-match or indels in adapter will reduce demultiplexing rate
3- quality of sequencing primers

TruSeq libraries sequenced using Illumina standard workflow can result in up to 4% undetermined reads plus the PhiX added

[zeam]

Thank you@nucacidhunter!

[pmiguel]

Usually when the percentage of undetermined is that high, it means the wrong index sequence was used. (Unless you had 50% phiX spiked in, as nucacidhunter said.) It is possible to determine if you have large numbers of unexpected indexes in your reads.

--
Phillip

[zeam]

Thank you, Phillip.