Salmon: library type disagreement

[ronaldrcutler]

Hi all,

I have been testing out salmon on RNA-seq data and am wondering what I should do about this output regarding library type. Basically, I set the expected library type to expect 'inward stranded reads coming from the reverse strand' (ISR), as this is what has worked in the past with the same data. More info on library types can be found here.

    
    "read_files": "( 3_TTAGGC_R1_001_merged.fastq.gz, 3_TTAGGC_R2_001_merged.fastq.gz )",
    "expected_format": "ISR",
    "compatible_fragment_ratio": 0.8452727256532248,
    "num_compatible_fragments": 38623083,
    "num_assigned_fragments": 45693043,
    "num_consistent_mappings": 39307792,
    "num_inconsistent_mappings": 10928813,
    "MSF": 0,
    "OSF": 14686,
    "ISF": 888461,
    "MSR": 0,
    "OSR": 51469,
    "ISR": 39307792,
    "SF": 1996372,
    "SR": 7961581,
    "MU": 0,
    "OU": 0,
    "IU": 0,
    "U": 0

Here most of the reads are agreeing with the expected library type, but some indicate that they come from the forward strand. Why is this? Is this an accurate reflection of the read data or is it dependent on how salmon looks at the reads?

-R