STAR ENCODE option

Question

[tianleivv]

Hi,

I am using STAR to align my RNA-seq data. I found in ENCODE options: the max number of multiple alignments allowed for a read is 20 (i.e. --outFilterMultimapNmax 20). Why it was not set as 1 to include only unique mapped reads for downstream analysis?

Thanks! Lei

Answer 1

[h.mon]

Because counting tools such as featureCounts and HTSeq can exclude multi-mapping reads, and other tools such as RSEM and eXpress can use the information from multi-mapped reads to provide more accurate count estimates.

[tianleivv]

Thanks h.mon! Why did you say RSEM can use the information from multi-mapped reads to provide more accurate count estimates? Do you have any references for this?

[h.mon]

RSEM paper. Read the section "Transcript quantification" from the review A survey of best practices for RNA-seq data analysis for a better explanation of the differences between aggregated raw counts (HTSeq and featureCounts) versus expectation-maximization estimated counts (RSEM, eXpress and Salmon).