Strand bias is very high

[thu1911]

Is the bug primarily related to salmon (bulk mode) or alevin (single-cell mode)?
salmon
Describe the bug
I used the default parameters to run salmon on mouse sample. But I got the warning:

Detected a *potential* strand bias > 1% in an unstranded protocol check the file:

To Reproduce
"expected_format": "IU", "compatible_fragment_ratio": 1.0, "num_compatible_fragments": 10869138, "num_assigned_fragments": 10869138, "num_frags_with_concordant_consistent_mappings": 10212463, "num_frags_with_inconsistent_or_orphan_mappings": 1088473, "strand_mapping_bias": 0.5258466052704426, "MSF": 0, "OSF": 0, "ISF": 4842274, "MSR": 0, "OSR": 0, "ISR": 5370189, "SF": 593930, "SR": 494543, "MU": 0, "OU": 0, "IU": 0, "U": 0

Specifically, please provide at least the following information:

• Which version of salmon was used?
  "salmon_version": "0.14.1"

• How was salmon installed (compiled, downloaded executable, through bioconda)? bioconda

• Which reference (e.g. transcriptome) was used?
  gencode mouse VM22

• Which read files were used?

Downloaded from sra（Just ordinary Illumina reads)

• Which which program options were used?

salmon quant -i $mm10_whole_index -l A \
    -1 $fastq1 \
    -2 $fastq2 \
    -p 20 --validateMappings -o $output_whole

Expected behavior
Why Strand bias is very high?

Desktop (please complete the following information):
Linux localhost.localdomain 3.10.0-957.el7.x86_64 #1 SMP Thu Nov 8 23:39:32 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux

[rob-p]

Hi @thu1911,

The strand bias here is actually quite moderate. The calibration of strand_bias is such that a value of 0.5 means there is no bias (i.e. half of the fragments start with read 1 on the forward strand and half start with read 1 on the reverse complement strand). Your value is "strand_mapping_bias": 0.5258466052704426, so you are pretty close to the ideal value of 0.5, though there is a slight bias in the data. I wouldn't be concerned about this assuming you were assuming an unstranded protocol.

Best,
Rob