map only unique reads

[Assa Yeroslaviz]

Hi,

I know this has probably asked a few times, but after trying to find a solution (not only in this forum) I am still not sure how to do it.

I would like to use STAR to map only unique reads (paired-end in my case).

With the term unique I mean, I would like to discard all duplications (multiple reads mapped to the same sequence), as well as all the read mapped to multiple loci. 

I have tried to  use the option --outFilterMultimapNmax set to 1. 

(I have pasted my complete command below)

I have tested the exact same command with and without this parameter. the results are exactly the same. I get the exact same amount of so-called Uniquely mapped reads. 

So my questions here are:

1. What exactly does this parameter is doing?

2. Is there a way to get only unique results from the STAR aligner, or do i need to run a de-duplication step afterwards?

thanks

Assa

STAR-STAR-2.5.1a/source/STAR  --runThreadN 15 --readFilesCommand zcat --genomeDir  ~/genomes/STARindexed/ --readFilesIn  ../rawData/$NEW_FILE\_R1.fastq.gz ../rawData/$NEW_FILE\_R2.fastq.gz  --outFileNamePrefix STARmapping/$NEW_FILE.rawData. --outFilterMultimapNmax 1  --alignIntronMax 1  --outSAMtype BAM SortedByCoordinate  --outWigType wiggle read1_5p  --outReadsUnmapped Fastx  --alignEndsType EndToEnd --limitBAMsortRAM 1003083170

[Alexander Dobin]

Hi Assa,

to output only one of the multi-mapping alignments, picked randomly out of the alignments with the highest score:

--outMultimapperOrder Random --outSAMmultNmax 1

On the other hand,  --outFilterMultimapNmax 1 simply reports only uniquely mapping reads, i.e. discards all the reads that are multi-mappers.

Cheers

Alex