Are there flags for applying these different strand deciding options from the manual? Am I missing something?
3.2 STRAND DECIDING METHOD
For non-stranded data, the original strand of the RNA molecule (fixed to be the predicted mainly expressed) must be determined. The following options differ by how the strand is decided (per region, per sample):
(Use only this option whenever good gene annotations are available) RefSeqThenMMSites - The strand is determined according to gene annotations (i.e. RefSeq annotations), by choosing the strand which is more likely expressed and sequenced (exon>>intron>>intergenic, try to solve conflicts according to known gene expression levels). If the annotation is unavailable or indecisive (such as exons on both strand), strand is decided according to the strand that fits the majority mismatch sites (e.g. if the number of A-to-G mismatch sites is higher- choose sense strand). This is the cleanest method and the only one present in the non-verbose output.
(verbose only) MMSitesThenRefSeq – The decision is made in the exact opposite order of the previous one – decide according the strand fitting the majority of the mismatch sites, and only if this is indecisive use annotations. Since this method chooses strand according to the mismatches sites, it maximizes noise as well, thus it is not recommended for general usage. One can estimate though how good the annotations were when comparing to the SNR of this output.
(verbose only) Randomly – This is mainly a negative control. Ideally, signal would be
half as much, noise still the same.
Are there flags for applying these different strand deciding options from the manual? Am I missing something?
3.2 STRAND DECIDING METHOD For non-stranded data, the original strand of the RNA molecule (fixed to be the predicted mainly expressed) must be determined. The following options differ by how the strand is decided (per region, per sample): (Use only this option whenever good gene annotations are available) RefSeqThenMMSites - The strand is determined according to gene annotations (i.e. RefSeq annotations), by choosing the strand which is more likely expressed and sequenced (exon>>intron>>intergenic, try to solve conflicts according to known gene expression levels). If the annotation is unavailable or indecisive (such as exons on both strand), strand is decided according to the strand that fits the majority mismatch sites (e.g. if the number of A-to-G mismatch sites is higher- choose sense strand). This is the cleanest method and the only one present in the non-verbose output. (verbose only) MMSitesThenRefSeq – The decision is made in the exact opposite order of the previous one – decide according the strand fitting the majority of the mismatch sites, and only if this is indecisive use annotations. Since this method chooses strand according to the mismatches sites, it maximizes noise as well, thus it is not recommended for general usage. One can estimate though how good the annotations were when comparing to the SNR of this output. (verbose only) Randomly – This is mainly a negative control. Ideally, signal would be half as much, noise still the same.