Often times it’s essential to know the strand orientation of your library. If for instance you have overlapping genes, or very closely nested genes, you wont be able to accuretly calculate transcription units unless you know the strand of origin of both your RNA library and your genes of interest.
In order to identify the strand orientation of RNA NGS libraries I
downloaded from SRA I stumpbled upon this excellent little tool called
infer_experiment.py from the RseqC. It’s
a great resource for this kind of thing. infer_experiment.py takes in a bam
file (aligned to your genome) and a BED file of genes with known strandedness.
So - to run it in parallel on all my bam libraries I just did a simple parallel command to expedite it as it’s quite quick.
parallel -j 2 "infer_experiment.py -r Zea_mays.AGPv4.36.gene_only_strand.bed -i {} > library_type_id_{/.}.txt" ::: ../all_aligned_files_sorted/*.bam
After this runs, the output looks like
This is PairEnd Data
Fraction of reads failed to determine: 0.0067
Fraction of reads explained by "1++,1--,2+-,2-+": 0.4973
Fraction of reads explained by "1+-,1-+,2++,2--": 0.4960
This is PairEnd Data
Fraction of reads failed to determine: 0.0104
Fraction of reads explained by "1++,1--,2+-,2-+": 0.5001
Fraction of reads explained by "1+-,1-+,2++,2--": 0.4895
So, what this output is telling you is that the reads are coming from either hte posotive or negative strand, and do NOT agree with the strandedness of the genes they’re aligning to. So this is an UNSTRANDED library.
In case you’re looking for some extra reading on strand libraries I found the following links extemely useful
https://chipster.csc.fi/manual/library-type-summary.html
http://onetipperday.sterding.com/2012/07/how-to-tell-which-library-type-to-use.html