SEQanswers

Go Back   SEQanswers > Bioinformatics > Bioinformatics



Similar Threads
Thread Thread Starter Forum Replies Last Post
STAR mapping error Him26 Bioinformatics 2 03-09-2016 04:17 PM
STAR Mapping taking too long mtiwaridros RNA Sequencing 1 05-02-2014 09:26 AM
STAR multi-mapping id0 Bioinformatics 1 03-28-2014 08:41 AM
cuffcompare output combined.gtf question arrchi Bioinformatics 0 09-01-2011 09:15 AM
cuffcompare: transcripts without strand and output combined alexnico84 Bioinformatics 0 01-13-2011 05:48 AM

Reply
 
Thread Tools
Old 12-20-2016, 11:11 AM   #1
Dani
Member
 
Location: Bay Area

Join Date: Dec 2016
Posts: 14
Default Weird output STAR mapping R1 and R2 separately vs combined

Hi,

I am just starting to learn using STAR for RNA-seq mapping. I made libraries from low-input, degraded (and likely enriched in short RNAs) samples. Because I ran my samples with someone else in lab, I did paired-end 2x150 bp read (kind of overkill, I know).
Before running STAR, I only took out low quality sequences and no adapter trimming since I read it does this, correct?
I am not too surprised by the low percentage unique mapped, but am confused about why the percent of short unmapped reads increases from 10% to 70% when I process both reads together (R1 and R2), while the multi mapped decreases from 50 to 10% - any help/ideas/suggestions?

R2
Time Speed Read Read Mapped Mapped Mapped Mapped Unmapped Unmapped Unmapped Unmapped
M/hr number length unique length MMrate multi multi+ MM short other
Dec 20 10:43:15 636.6 10610538 95 34.0% 102.7 0.7% 50.9% 4.8% 0.0% 10.4% 0.0%
Dec 20 10:44:15 559.1 18636117 94 33.7% 101.5 0.7% 51.1% 4.8% 0.0% 10.4% 0.0%
Dec 20 10:45:30 552.9 29951091 94 33.7% 101.3 0.7% 51.2% 4.8% 0.0% 10.4% 0.0%

R1
Time Speed Read Read Mapped Mapped Mapped Mapped Unmapped Unmapped Unmapped Unmapped
M/hr number length unique length MMrate multi multi+ MM short other
Dec 20 10:35:34 608.2 10473886 97 35.2% 105.2 0.6% 49.0% 4.6% 0.0% 11.2% 0.0%
Dec 20 10:36:34 679.8 23039300 96 34.9% 104.1 0.6% 49.2% 4.6% 0.0% 11.3% 0.0%
Dec 20 10:37:34 691.4 34952409 96 34.9% 104.4 0.6% 49.2% 4.6% 0.0% 11.3% 0.0%

R1 and R2
Time Speed Read Read Mapped Mapped Mapped Mapped Unmapped Unmapped Unmapped Unmapped
M/hr number length unique length MMrate multi multi+ MM short other
Dec 20 09:23:11 346.1 5960173 194 13.8% 168.1 0.7% 11.7% 3.9% 0.0% 70.7% 0.0%
Dec 20 09:24:13 386.1 13297340 192 13.7% 167.1 0.7% 11.8% 3.9% 0.0% 70.6% 0.0%
Dec 20 09:25:15 400.8 20709447 191 13.6% 166.0 0.7% 11.9% 4.0% 0.0% 70.5% 0.0%
Dec 20 09:26:15 404.5 27640580 191 13.6% 166.2 0.7% 11.8% 4.0% 0.0% 70.6% 0.0%
Dec 20 09:27:15 402.5 34210699 191 13.6% 165.9 0.7% 11.8% 4.0% 0.0% 70.6% 0.0%
Dani is offline   Reply With Quote
Old 12-20-2016, 12:20 PM   #2
Brian Bushnell
Super Moderator
 
Location: Walnut Creek, CA

Join Date: Jan 2014
Posts: 2,707
Default

I highly recommend doing adapter-trimming before alignment, especially with small RNAs.

I think 2x150bp reads in which you expect full overlap plus overhanging adapter sequence for all reads will probably confuse many aligners. You might want to merge them and just map with the consensus as single-ended.
Brian Bushnell is offline   Reply With Quote
Old 12-21-2016, 07:00 PM   #3
Dario1984
Senior Member
 
Location: Sydney, Australia

Join Date: Jun 2011
Posts: 163
Default

STAR doesn't do adapter trimming, but it does do soft-clipping, which gives the same effect. I also have a dataset where the pairs of reads largely overlap, so I also need to merge the pairs into a consensus sequence before I can use STAR. I read on the STAR forum that a future version of the software will handle these kinds of datasets correctly.
Dario1984 is offline   Reply With Quote
Old 12-21-2016, 07:27 PM   #4
Brian Bushnell
Super Moderator
 
Location: Walnut Creek, CA

Join Date: Jan 2014
Posts: 2,707
Default

Soft-clipping is not equivalent to adapter-trimming. And, generally, I view automatic soft-clipping as a negative when evaluating aligners - it basically means the aligner is not confident of providing a correct alignment at that location, so it would rather pretend that location does not exist, which incurs severe ref-bias. Relying on soft-clipping to solve library-creation problems will hide a lot of real mutations. Your alignments will be more accurate if you first trim adapters, then map the reads. It ends up being faster, as well. For overlapping reads, though, I do recommend merging prior to mapping; with a good merging tool, that will also eliminate some of the adapter sequence, lead to more accurate quality scores, and also give increased confidence in variant calls, since the farther a variant is from the ends of a read, the greater the confidence.

Last edited by Brian Bushnell; 12-21-2016 at 07:38 PM.
Brian Bushnell is offline   Reply With Quote
Old 12-30-2016, 12:57 PM   #5
Dani
Member
 
Location: Bay Area

Join Date: Dec 2016
Posts: 14
Default

Thanks, I will try that!
Dani is offline   Reply With Quote
Reply

Tags
rna-seq, short unmapped reads, star

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off




All times are GMT -8. The time now is 04:39 AM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2018, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO