Tricks And Tips For Ribosome Profiling?

Question

[Zev.Kronenberg]

Greetings all,

I have just started collaboration on a ribosome profiling project. We have ~30bp illumina single end reads (adaptor removed). These sorts of data are new to me and I am looking for some advice on prudent and well established analyses.

Here are my questions:

What are people using to align short RNA sequences?

I was thinking Bowtie 2 or BWA.

Should I toss out tRNAs and rRNAs prior to the alignment step?

One PI said yes, but I am not convinced. The tRNA and rRNA are in the reference genome and I figured I could mask them out later.

What bioinformatic controls are there?

I was going to look at some house keeping genes, but don't know of a metric to use.

Besides my experiment specific questions, what analyses are commonplace?

Any must read papers?

Answer 1

[Gareth Morgan]

Hi Zev. I am not an expert at this, but I'm trying to learn how to do the same sorts of analyses so I'm in a similar position. Although I'm slightly distracted by trying to to the wet work as well.

The aligner that you use will depend on your organism. I work on E. coli and use Bowtie 0.12.8, but Bowtie 2 might deal better with splice junctions. I use don't pre-filter the reads for non-mRNA, again because in E. coli there are multiple rRNA loci so throwing out any reads that map to more than one location gets rid of them.

Controls would include looking for whether your footprint reads align with regions you'd expect to be translated, and whether you see high occupancy at known pause sites. Again, it depends on your organism.

The standard analysis to do is translation efficiency: how many ribosomes are translating each message. You need footprints and total mRNA samples in order to calculate this for each gene. Beyond that you can look at pausing patterns and identify novel ORFs and non-cannonical start and end sites.

As for papers, you'll probably want to have a look at Nick Ingolia and Jonathan Weissman's recent publications. There's quite a lot.

http://www.ingolia-lab.org/publications.html

http://weissmanlab.ucsf.edu/publications/publications.html

There's a discussion of data handling here: http://www.ncbi.nlm.nih.gov/pubmed/20946809 and this paper talks about the kinds of analysis that are possible: http://www.ncbi.nlm.nih.gov/pubmed/22056041

Answer 2

[JC]

What are people using to align short RNA sequences?

You will face the same problem than RNAseq data, some portion of short sequences will map correctly to one or more positions, others will have errors, and some will be in the middle of a splice site. Personally I like Bowtie/BWA for a quick alignment, but Blat can be more sensitive (and really slow compared with the others). For example, in a common RNAseq library (Illumina GAII, 75bp, se) I can map ~70% of the reads to the human reference with Tophat/Bowtie/BWA, with 60% of reads with multihit. In contrast, Blat can map ~95% of the reads with less than 30% multihit reads (I postprocess the hits filtering the best hit with a UCSC-like score).

Should I toss out tRNAs and rRNAs prior to the alignment step?

Depend on the proportion of expected reads, when you are selecting your fragments without polyA selection, rRNAs are highly dominant in the sample (20-60% of the reads). I remove them doing a quick alignment to ribosomal sequences before mapping to the reference, this simple and fast step saves computing time (10-30% of the time)

What bioinformatic controls are there?

I cannot think in a well defined control.

Besides my experiment specific questions, what analyses are commonplace?

Check the reference below.

Any must read papers?

Definitively http://www.cell.com/abstract/S0092-8674(11)01192-5