a strange peak in LncRNA library

[chenyudr]

Hi, all,
We met a strange peak in prepared LncRNA library. The library construction procedure is listed below.
1 micro gram total RNA after Ribozero treatment was used at starting material and ribosomal RNA was almost completely removed as PicoRNA chip map indicated and Ribozero- RNA is ~20-50ng. Library construction was performed following NEBNext® Ultra™ RNA Library Prep Kit for Illumina protocol with size selection for 300-400bp fragments. The PCR cycle number is ~15 cycle.
In 12 of 16 samples, we found a strange sharp peak, which unlike those we usually obtained.
Did you ever meet similar results? anyone could give us an explanation?
Thanks!
attachment 1: Strange peak in library 2100 map
attachment 2: RNA after Ribozero treatment in Pico RNA map.

[nucacidhunter]

It seems that you have used a gel based size selection method. If this is the case, peak is normal.

[chenyudr]

Thanks for your reply, we used beads for size selection.
Does the library could be sequenced and produce normal data?

[nucacidhunter]

The kit that you have mentioned does not seem to include a size selection step and size selection of RNA-Seq libraries is uncommon. The library can be sequenced but data analysis will be difficult because somehow you have to account for the biases introduced by size selection that does not seem have worked well.

[cmbetts]

Did you reduce the RNA fragmentation time to have longer fragments at the time of cDNA synthesis? It does look like the trace of a standard protocol RNA-Seq library that has had the lower MW fragments hacked off by a low ratio Ampure selection.
Here's a thread about the Clontech Stranded kit that has a slide showing the effects of different RNA fragmentation times on the final library profile http://seqanswers.com/forums/showthread.php?t=44453. I believe the two kits have different divalent cation concentrations for the fragmentation, so you probably can't just lift the conditions from the slide, but the same principle would apply if you can afford the time/kits to do some empirical testing (If having this larger size library is actually important, which is probably only true if you're actually trying to assemble rather than count).