strandedness of rnaseq experiments

[batman]

Hi,
I am trying to generate some statistics with picard CollectRnaSeqMetrics utility. The parameter STRAND_SPECIFICITY requires a value described as follows.
----------------------------------
For strand-specific library prep. For unpaired reads, use FIRST_READ_TRANSCRIPTION_STRAND if the reads are expected to be on the transcription strand. Required. Possible values: {NONE, FIRST_READ_TRANSCRIPTION_STRAND, SECOND_READ_TRANSCRIPTION_STRAND}
---------------------------------
my library is a paired library, so do I need to use value NONE? otherwise, can you tell me how to determine the correct value?

[swbarnes2]

Hi,
I am trying to generate some statistics with picard CollectRnaSeqMetrics utility. The parameter STRAND_SPECIFICITY requires a value described as follows.
----------------------------------
For strand-specific library prep. For unpaired reads, use FIRST_READ_TRANSCRIPTION_STRAND if the reads are expected to be on the transcription strand. Required. Possible values: {NONE, FIRST_READ_TRANSCRIPTION_STRAND, SECOND_READ_TRANSCRIPTION_STRAND}
---------------------------------
my library is a paired library, so do I need to use value NONE? otherwise, can you tell me how to determine the correct value?

Unless you know otherwise, assume that the library prep was unstranded. So for some of your reads, read 1 will be in the direction of the transcript, and in some, it will be read 2.

So say None.

[batman]

Got it! Thanks a lot...

[dvanic]

If you want to test whether you were lucky enough to get given a strand-specific library you can use the infer_experiment.py module of rseqc.

http://rseqc.sourceforge.net/

rseqc has a whole bunch of other very useful tools to do various QC on RNA-seq data...

[batman]

Thanks dvanic... thats a swissknife of rnaseq utilities..