Quality score documentation

[kaizhang]

It is currently unclear whether the quality scores return by a given function has been adjusted. For example, for a noodles::sam::Record, .quality_scores().iter() and .quality_scores().as_ref() give different outputs. The former has been offset by 33, while the latter contains the raw data. It is also unclear if the quals are adjusted when converting to Bam record. What should I expect when calling .quality_scores().as_ref() with a Bam record?

[zaeleus]

When dealing with a format record, the AsRef<[u8]> implementation does return the underlying raw data from the record. In sam::record::QualityScores, these are the ASCII encoded scores (score + offset); in bam::record::QualityScores, the raw phred-scaled scores. They are simply a wrapped view of the raw field.

The implementations are there for more advance usages, in the case the raw data is needed. However, the typical usage is through the sam::alignment::record::QualityScores implementation, which guarantees normalization of the scores.