Skip to content
Omnikalc

Phred quality score calculator: how many errors are in your sequencing data?

Enter a Phred quality score, the read length and the number of reads to get the error probability, the accuracy, the errors per read and the share of error-free reads.

Updated on
Your inputs

Value out of range

bases

Value out of range

million

Value out of range

Results

Probability that a base call is wrong

0.1000 %

Below Q20: unreliable base calls
Base call accuracy
99.9000 %
One error expected every
1,000 bases
Expected errors per read
0.150
Reads with no error at all
86.1 %
Wrong bases in the whole run
60,000,000

DNA sequencers give each base a quality score on the Phred scale. This Phred quality score calculator turns the score Q into the error probability, the accuracy, the expected errors per read and the share of reads with no error. With the number of reads it also gives the total number of wrong bases in the run.

How the calculation works

  • Enter the Phred quality score Q of the bases, the read length and the number of reads.
  • The probability that a base call is wrong is P = 10^(−Q/10), and the accuracy is 1 − P.
  • Q10 means 1 error in 10, Q20 means 1 in 100, Q30 means 1 in 1,000 and Q40 means 1 in 10,000.
  • The calculator gives the expected errors per read, the share of reads with no error (1 − P)^length and the total number of wrong bases in the run.

Example

A sequencing run has Q30 bases, reads of 150 bases and 400 million reads. The error probability is 0.1 %, so one base in 1,000 is wrong. Each read has 0.15 errors on average, 86.1 % of the reads have no error, and the run holds 60 million wrong bases among its 60 billion.

Why it matters

Quality scores are used to filter and trim reads, to judge a sequencing run, to choose parameters for alignment and variant calling, and to understand why some applications need a higher quality. The calculator turns the logarithmic scale into numbers that are easy to picture.

Good to know

The score is the platform’s own estimate and not a measured error rate, and the calculator assumes the same Q for all bases and independent errors. In real data, the quality changes along the read and between reads, and errors are not independent. For a mixture of scores, calculate the mean error probability, not the mean Q. Phred scores in FASTQ are stored as characters with an offset of 33.

The DNA concentration calculator covers sample preparation, the DNA copy number calculator covers the number of template molecules, the log reduction calculator uses the same logarithmic scale, and the binomial calculator covers the probability of a number of errors.

Frequently asked questions

What is a Phred quality score?

It is a number Q = −10 log₁₀(P) that a sequencer attaches to each base call, where P is the estimated probability that the call is wrong. It was created for the Phred program of the Human Genome Project to read the chromatograms of Sanger sequencing, and is now the standard in the FASTQ format used by Illumina and other platforms.

What do Q20 and Q30 mean?

Q20 means a 1 % chance of error, that is an accuracy of 99 %. Q30 means 0.1 %, or 99.9 % accuracy. Q40 is 0.01 %. Each step of 10 in Q is a reduction of the error probability by a factor of 10. Sequencing providers often report the percentage of bases at or above Q30.

Why does the error-free share of reads fall with length?

The errors add up along a read: a read of L bases with an error probability P per base has no error with a probability (1 − P)^L. For Q30 and 150 bases, this is 86 %, but for Q20 and 150 bases it is only 22 %, which is why a high Q is important for long reads or for the detection of rare variants.

Is the score an exact error rate?

It is an estimate based on a model of the signal, and a well calibrated platform gives a true error rate close to it. Real error rates also depend on the position in the read, the sequence context and the chemistry. Errors can be larger than predicted, so some pipelines recalibrate the scores with known sites.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.