This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Pre-processing duplicates in SMARTer-seq data

Hi, I have reads generated from FFPE samples by using SMART-Seq Total RNA Pico Input with UMIs (ZapR® Mammalian) kit. We recieved demultiplexed files from the sequencing facility.

First, we preformed read trimming and deduplication with fastp (fastp -c --trim_poly_x –dedup), and used FastQC for additional visualization of pre-processing (especially for duplicates). What is confusing me is that the number of deduplicated reads before and afrer using fastp is approximately the same. Moreover, the number of duplicates in R1 is much higher than the number of duplicates in R2 in each sample (which can be explained by the fact that R2 contains UMI sequence, according to the man).

I have a few questions about this matter:

  1. Does the number of duplicates pre- and post- fastp processing appear almost unchanged because fastp works in paired-end mode, but FastQC analyzes each fastq file separately? So if two reads have identical sequence in R1 but differ in R2 (e.g. different UMIs), they will not be marked as duplicates by fastp - but FastQC still marks the two R1 sequences as identical.
  2. Would it be more approprate to use fastp --umi instead of fastp --dedupl? by having read this post (suggesting to use tool tally to de-duplicate identical fastq record, because fragmenatation is happening before PCR in librabry prep), then using --dedup makes sense as it does the same thing, if I'm not wrong.
  3. Should I continue the analysis as usual, following the standard steps, or you have any suggestion to modify the pre-processing?

Any advice and suggestion would be appreciated!

rna-seq umi fastq smarter-seq fastp
• 990 views • link
updated by ATpoint ATpoint • written by kiki • kiki

In the manual you linked TakaraBio recommends using their software called Cogent. CogentAP

collapses UMIs and trims 3 nucleotides of UMI linker and 3 nucleotides derived from the Pico v3 SMART UMI Adapter

If you are using your own method then make sure you are doing the following when analyzing the results:

• Read 1 matches the antisense sequence of the input RNA.
• Read 2 corresponds to the sense strand.
• First eight cycles of Read 2 belong to UMIs followed by 3 nucleotides of UMI-linker and 3 nucleotides derived from the Pico v3 SMART UMI Adapter.
• Trim 8 nt UMIs + 3 nt UMI linker + 3 nt from the SMART UMI Adapter from Read2 prior to mapping
• 0 views • link
written by GenoMax GenoMax

I am new to RNA-seq so I was confused about hard-trimming the first 14 nt from the Read 2, as I don't know the amount of the sequencing noise present. So doing something like

fastp \
  --umi \
  --umi_loc=read2 \
  --umi_len=8 \
  --trim_front2=14 \
  --trim_poly_x \
  -c

, then mapping the reads and doing the deduplication after the mapping would be appropirate?

• 0 views • link
written by kiki • kiki

There should be no sequencing noise. Follow the kit directions and hard frim first 14 bases. Are you going to account for the UMI first?

• 0 views • link
written by GenoMax GenoMax

Okay, thank you for the advice. Yes, we are going to account for the UMIs first.

• 0 views • link
written by kiki • kiki

Don't delete content that received community help. It's a knowledge repository after all.

• 0 views • link
written by ATpoint ATpoint

0 answers

No answers yet.

Log in to answer this question.

More posts like this