beginner problem: coding protein-coding genes in multi-gene alignment

[lcfcrespo]

Hello all,

I am a beginner user learning how to use IQtree for simple ML analyses of 2 gene-alignments (invertebrate mitochondrial COI + ribosomal 28 S) and am having some trouble coding the partition for the COI.
By ordering IQTree to analyze each reading frame of the COI as I did before for other softwares, it does the translation and immediately reports the presence of stop codons, thus ending my analysis.
I used the following partition file:

#nexus
begin sets;
charset part1 = CODON5, 1-657\3;
charset part2 = CODON5, 2-657\3;
charset part3 = CODON5, 3-657\3;
charset part4 = DNA, 659-1399;
end;

I attach the output of this failed run
parts.txt.log

I understand the solution must be something simple, but I can't find what it is from the tutorials, since I inform IQTree that my COI sequences must be read with the invertebrate mitochondrial genetic code and I add all 3 reading frames of the gene, and had it sized to be multiple of 3 (original alignment size was 658, I cut one position to be readable as Codon).
Would be most appreciated for any help and sorry to bother with such a basic problem.
Luís
P.S.: I tried to post this on friday, as my first post after making my account, but then the account suddenly got suspended for some security check from GitHub, so I don't know if my original message got through and is being moderated by someone. If that is the case, then it's best that this message gets deleted to avoid spamming.

[bqminh]

The reason is due to CODON5 datatype. Please change all CODON5 to DNA and it should work.

#nexus
begin sets;
charset part1 = DNA, 1-657\3;
charset part2 = DNA, 2-657\3;
charset part3 = DNA, 3-657\3;
charset part4 = DNA, 659-1399;
end;

If you want to know the reason: For CODON data types IQ-TREE will use codon models, which need the full coding sequence in the partition to convert every tri-nucleotide into non-stop codon states and apply a 61x61 substitution matrix for the analysis.
However, you have partitioned the data into codon positions, each will be modelled using a DNA substitution model. Therefore you have to use DNA datatype for all partitions.

Note: I don't know why your partitions 1-3 ends at 657? Probably a typo? If you write this, then site 658 will be ignored from the analysis.

[lcfcrespo]

Many thanks for the feedback.
I ended the COI at 657 to make it divisible by 3. I see now that was a mistake.
But then, at any point in the command is unnecessary to state to IQTree that the positions 1 to 658 is indeed a protein-coding gene (using the invertebrate mitochondrial genetic code)...?
My comprehension of the tutorial was insufficient, I see it now. Again, many thanks for the help.