FastQ ASCII table for raw sequences' QC

usegalaxy.org support
,
1

Hello everybody,

I’ve been reading many tutorials and educational materials about basic .fastq files quality control procedures but I’m confused because it doesn’t seem to apply to my experimental data. I was able to upload some .ab1 files from sanger semi-automated sequences to the Galaxy platform and to convert them into .fastq files using https://www.usegalaxy.org.au

Until now any problems. In fact, that’s where they begin, because after retrieving the results I realized something weird about them because it seems they included symbols from 2 different ASCII tables used for this purpose: at the same time my files contain symbols like the characters “#” and “%”, which belong to the 33-base ASCII table, they also contain uppercase letters of the end of the alphabet (R, S, T, U, V, W, X, Y and Z), which belong to the 64-base ASCII table only.

Please note I’m taking as reference the ASCII quality scores corresponding to the tables available at Quality (Phred) scores .

I’m confused! If .fastq sequence scores go from 1 (or zero) to 40, how can it be that I have a quality score composed of 64 characters?

Any clarification is greatly appreciated.

P.S.: some time after having posted I remembered to have seen somewhere the option to share my Galaxy history with other people, an option I didn’t find anymore after the question was posted.

2

Hi @lcfiorini
in the Phred+33 format the range of allowed ASCII values extends from 33 (= !, the first non-whitespace printable character of the ASCII table) all the way to 126 (= ~, its last printable character). This means the encodable Phred quality scores range from 0 to 93.
Everything beyond that are just machine and base caller-specific conventions.
Phred+64 is of historical interest only nowadays (Phred+33, also called fastq Sanger format, is the de facto standard). Because it used ASCII value 64 (= @) as its lowest value, but, like Phred+33, has to end at 126 for printability, it can represent only a smaller range of quality scores.
At the time, when Phred+33 and Phred+64 were both in active use, their ranges would normally not overlap much because base calls from sequencers at the time would have relative large error probabilities, but Phred+33 scores not extending (or not too much) into the Phred+64 range was never a format specification (just a convention at best).
Modern sequencers can for sure produce base quality scores of more than 40 and so what you’re seeing is not at all surprising.

3

The last paragraph of the page you linked to in your question, “Recognizing the format”, tries to explain what I just said, too.