[nilshomer]

According to SAM manual, the column 5 is for mapping quality. I just refer to this column as mapping quality, and I think the two - "alignment score" and "mapping quality" - are the same.

Alignment score and mapping quality are not the same. The former is a measurement of the distance (or log-odds see:BLOSUM) between the observed sequence and reference sequence while the latter is generally the Phred-scaled probability of mismapping. In the SAM format, the alignment score can be stored in the "AS" tag while the fifth column is the mapping quality.

[clariet]

Thank you very much for the reply. I was referring to the 5th column. It IS the mapping quality according to SAM manual. But I guess this mapping quality must be correlated with alignment score for some way.

I have seen a lot of alignment with low mapping quality (0 or 1). For the input of cufflinks, do you usually filter out these low quality mapping reads? what is the cutoff you usually use?

Thanks

Hi there,

That is actually not the score, but the mapping quality (I'm assuming you're referring to column 5 for 255 "score"). For calculation of score, you will have to take the alignment in colorspace (XL:Z) and the number of colorspace mismatches (XU:Z), then use SOLiD's formula.

A score of 10 shouldn't be appearing in the output. The seed of 25bp mapping with at most 2 mismatches will give u the lowest possible score for an alignment to be reported, which in my case is 18.

[Haneko]

Hi clariet,

Actually I've never considered this column to be a factor in filtering, so i can't give you a definite answer. But I think it'd be good to filter away those low quality mappings, perhaps someone could give a recommendation on the threshold?

[haolili]

I am also ready for analysis of AB solid data with TopHat and later Cufflinks, hope for successful example here