Bio.codonalign.codonseq module (cal_dn_ds) for SARS-COV-2

Question

I have calculated the dN,dS ratio for a SARS-COV-2 data set ('NC_045512.2' (Wuhan strain) and 'LC666924.1'), and results in a zero value (for most of the genes).

Is it possible and what is the interpretation, for example negative selection?

Answer 1

Yes that is the expected result, because there are no amino acid mutations in the data sets you are using. Thus the numerator is 0 and anything divided by zero is always zero, even if every codon had a synonymous mutation.

The most precise interpretation is simply there are no amino acid mutations. It doesn't matter how you reword that, all divergence is resulting from synonymous mutations.

The only thing to watch out for is a rounding error, i.e. a single amino acid change is dropping off detection because dN/dS is incorrectly rounded.

---

A post was raised (now deleted) about what does that say about selection. dN/dS theory states that if it is significantly >1 thats positive selection .. etc. However, a truer representation of the data is there is no amino acid mutation, rather than their is a significant excess of synonymous to non-synonymous mutations. A zero value makes things tricky, and is exemplified by two hypothetical data sets one where every codon had one synoymous mutation (strong negative selection) compared against one that had a single synonymous mutation (little mutation) and both data sets had zero amino acid mutations - both data sets would have a dN/dS of zero but that is not a reflection of negative selection.

---

To answer the point below a single mutation is not a great basis draw statistical significance and the dN/dS ratio is not a good test if either the numerator or denominator are zero.

Most data is under negative selection in strict accordance with dN/dS.

Neutral selection is extremely rare.

Summary Sometimes the statistical analysis of a data set does not say very much ... and sometimes we have to accept that as an answer.

Answer 2

So if that had a little mutation, Does that mean it's a neutral selection? And if the sequences (codons and amino) are exactly the same? We can say something about the interpretation?

See examples below-

ORF8 Seq1: MKFLVFLGIITTVAAFHQECSLQSCTQHQPYVVDDPCPIHFYSKWYIRVGARKSAPLIELCVDEAGSKSPIQYIDIGNYTVSCLPFTINCQEPKLGSLVVRCSFYEDFLEYHDVRVVLDFI Seq2: MKFLVFLGIITTVAAFHQECSLQSCTQHQPYVVDDPCPIHFYSKWYIRVGARKSAPLIELCVDEAGSKSPIQYIDIGNYTVSCLPFTINCQEPKLGSLVVRCSFYEDFLEYHDVRVVL--I

ORF10 Seq1 MGYINVFAFPFTIYSLLLCRMNSRNYIAQVDVVNFNLT Seq2 MGYINVFAFPFTIYSLLLCRMNSRNYIAQVDVVNFNLT

E (The same) Seq1 MYSFVSEETGTLIVNSVLLFLAFVVFLLVTLAILTALRLCAYCCNIVNVSLVKPSFYVYSRVKNLNSSRVPDLLV Seq2 MYSFVSEETGTLIVNSVLLFLAFVVFLLVTLAILTALRLCAYCCNIVNVSLVKPSFYVYSRVKNLNSSRVPDLLV