Is MAPQ score related to Phred quality score?

Question

[dpc]

Hi, I just wanted to know although the two quality scores have different formulae to calculate them, are they related to each other? I mean, can I say good MAPQ score implies the base qualities of the read is good?

Thanks, DC7

[GenoMax]

Take a look at this blog post for information about MAPQ and its implementation by various software programs.

MAPping Quality. It equals -10 log10 Pr {mapping position is wrong}, rounded to the nearest integer. A value 255 indicates that the mapping quality is not available.

Answer 1

[Jeremy Leipzig]

Phred is 10 log10P where in Phred P is the probability of a miscalled base.

Samtools doesn't really support bit-scores or e-values akin to what you would see in a BLAST report. There is no sensical formula for mapping quality, since some reads map to multiple locations, others align poorly but that can be because of the read itself or SNPs. As the blog post shows, every tool rolls its own MAPQ.

[dpc]

I am working with shotgun metagenome sequences with MetaPhlAn. It uses bowtie2 aligner and discards reads with MAPQ value less than 30. So, does it mean it proceeds only with the reads having good quality bases? Also, does it mean I should not do QC check step before? (provided, I am working only with already submitted data sets in SRA)

Thanks, DC7

[Jeremy Leipzig]

No Phred refers to bases and MAPQ refers to the alignment. You can do a QC step such as filtering and trimming reads, but that practice is less popular than it used to be.

[dpc]

Yes sir, I have done this step on some already submitted data. And I've got some strange result. I have checked with fastqc followed on 78 samples followed by multiqc run (which just accumulate the data into single file). Here's the result I've got. Can you please tell me whether I should trim bases less than phred quality score 20?

[Jeremy Leipzig]

You can certainly try that but it might actually make your results worse. Most aligners are already quality-aware.