Help with UMI (unique molecular identifiers) data processing

[donquijotes]

I've been browsing different papers and publications and trying to figure out what's the best way to analyze data with UMIs.
So far I have used GATK to do some analysis couple of times, but other than that I was mostly playing with alternative splicing analysis so I'm rather new to this CNV calling with UMIs topic and area of research.
What I would like to do is have the following design.

adaptor-UMI-DNAlibraryINSERT-UMI-adaptor

The UMIs will be 5 random bases on each side.

I get the whole UMI thinking and analysis but what I haven't found yet is the software to do such analysis. I've seen few tools to mark/find UMIs and put them on the header of the fastq sequence but then what? How do you bin and get rid of the true PCR duplicates? Does Picard have a function for it? If I have to write my own code then I'm out of luck lol.

I know Agilent supports UMIs with their Haloplex HS kits and their Surecall software that is mostly (from what I've heard) a nice GATK GUI.

Any help and guidance would be much appreciated. Newbies have the right to learn too, right?

Thank you in advance

[nucacidhunter]

Product described in web page below uses Molecular Indexing and the sequences are given in product manual.

http://www.biooscientific.com/Next-Gen-Sequencing/Illumina-RNA-Seq-Library-Prep-Kits/NEXTflex-qRNA-Seq-Kit

They have described analysis step in a link in page below:

http://www.biooscientific.com/Next-Gen-Sequencing/Illumina-RNA-Seq-Library-Prep-Kits/NEXTflex-qRNA-Seq-Kit

[charlescoldroom]

I am also interested to know about how to handle UMIs and remove duplicated reads based on UMIs.

I am using modified primers to have amplicon pools.

Which tools are there to mark/find UMIs and put them on the header of the fastq sequence? How could I then process the reads?

I have tried looking around, but I could not find any good step-by-step explanation, even papers just mention that they do the analysis but do not explain how.

Thanks!

[danwiththeplan]

Molecular indexing

Hi, you could try the script mentioned here:

http://www.biooscientific.com/Portals/0/White%20Papers/qRNA-Analysis.pdf

It's currently not working for me, but I'm in communication with the maintainer so I'll repost if I get everything working.

[luc]

A very simple approach would be to do a general de-duplification of the reads with BBTools (I have not used it for thispurpose but it should be better than our in house script) which will likely require a considerable memory. Then you should trim the 5 random bases.

[charlescoldroom]

Thanks guys, I will check out the suggestions!

[IanSudbery]

I know this most is a few months old now, but you might like to try our UMI-tools package, which offers a range different algorithms for deduplicating UMI sequences.

https://github.com/CGATOxford/UMI-tools

[danwiththeplan]

I know this most is a few months old now, but you might like to try our UMI-tools package, which offers a range different algorithms for deduplicating UMI sequences.

https://github.com/CGATOxford/UMI-tools

Hi, thanks for this contribution..

I'm reading the code, and this is what it looks like to me, but am I correct in saying that this script would correctly deduplicate splice-aware mappings ? i.e. reads that jump across splice boundaries are handled correctly?

[sudders]

Hi, thanks for this contribution..

I'm reading the code, and this is what it looks like to me, but am I correct in saying that this script would correctly deduplicate splice-aware mappings ? i.e. reads that jump across splice boundaries are handled correctly?

You've probably worked this out already, but yes, it handles splice-aware mappings.

[medalofhonour]

This group recently published a paper with a pipeline for analyzing UMI datasets. The software can be found here :

https://github.com/mikessh/mageri

[cement_head]

If you are using CLC Genomics Workbench:

https://www.qiagenbioinformatics.com/plugins/molecular-indexing

[Strandlife]

You should try Strand NGS for UMI protocols.
Strand NGS is the only software to provide comprehensive and end-to-end support for multi Unique Molecular Identifier Protocols

Few features includes:

1. Protocol diversity. Strand NGS supports data analysis from UMI protocols
i. Qiagen GeneRead®
ii. Archer VariantPlex®
iii. Rubicon Thruplex®
iv. Bioo Scientific NextFlex®)
v. A robust interface to specify custom UMIs

2. End-to-end or point-to-point. Users can go from reads to variants, can start at aligned BAMs containing the BC tag, or start/end at any reasonable point in the alignment/analysis workflow.

3. Workflow diversity. Strand NGS supports UMI protocols in DNA-, RNA- and small RNA-Seq workflows

4. Somatic- and UMI-ready visualizations. The genome browser visualizes consensus read lists. Each read contains UMI-related metadata, such as family size, UMI and mate UMI. A filter allows the easy exclusion of wild-type reads. This is useful at high sequencing depths and low allele frequencies, typical of data from somatic/tumor samples.

You could get a 20-day free trial by registering here with your organization email id:

http://www.strand-ngs.com/signup/freetrial

[[email protected]]

You can use fastp to preprocess UMI from fastq.