Genome annotation

Question

I have helped a lab sequence a mitochondrial genome. I used then MITOS to annotate the genome (warning them that I did it just for curiosity as I am not experienced). They submitted the sequence and annotation to genbank but it was refused for different reasons.

It's the genome of a species for which there is no reference. It's a jumping spider and the genome is almost identical to the one below. I used MITOS web server and I got an annotation almost identical to this one (http://mitos.bioinf.uni-leipzig.de/result.py?hash=bJuyJw4b) which corresponds to this sequence (https://www.ncbi.nlm.nih.gov/nuccore/MH891570 ). But the submitted annotation to Genbank is slightly different so I guess they manually edited it or they used another software. I am not sure I can share more specific information without the permission of the lab.

You can download a GFF file or the raw sequence at the page in the link (for the similar sequence).

My questions are as follows:

• after applying MITOS, is there any manual or automatic review that should be applied?
• Is it common that the annotation returned by MITOS is not good enough?
• Or could that indicate that maybe there is a problem in the assembled genome? (looks unlikely except for a few SNPs as three assemblers return almost the same sequence)
• Should we edit the annotation ourselves or there is a software that does it itself?
• How often using MITOS is enough or how often manually editing is needed?

The sequence corresponds to an arachnid. The reasons given by Genbank are:

Some or all of the sequences contain incorrect annotation and/or contain little similarity to other sequences in the database based on BLAST searches. Please also review your nucleotide sequences for issues such as low quality data or frameshifts.

Incorrect annotation is:

• coding regions lack stop codons
• CDS features do not have the correct nucleotide spans or location for example, nad5 is longer at the amino terminus compared to similar proteins in the database and extends into the flanking tRNA.

The first part of their reply is apparently generic as I have seen the same answer in another submission in which I didn't participate. I am not sure it's even verified as there are quite good BLAST matches. I have no idea about the other part. I just applied MITOS.

I used the invertebrate mitochondrial translation table option in MITOS. I guess it's the correct one as the result quite makes sense except for small adjustments.

Answer 1

Any regions annotated as protein sequences must have a complete protein sequence, including start and stop codons. It's somewhat common for mitochondrial genomes to require polyA extension for the stop codons to be completed; I expect that NCBI is asking that the submissions make this explicit. See, for example, our lab's Nippostrongylus brasiliensis mtDNA submission:

https://www.ncbi.nlm.nih.gov/nuccore/NC_033886.1

As an example in this genome annotation, the ND5 gene requires extension, as mentioned in the note:

3858..5439
/gene="ND5"
/locus_tag="B2K98_mgp10"
/note="TAA stop codon is completed by the addition of 3' A
residues to the mRNA"
/codon_start=1
/transl_except=(pos:5439,aa:TERM)
/transl_table=5
/product="NADH dehydrogenase subunit 5"
/protein_id="YP_009347353.1"
/db_xref="GeneID:31081492"

It doesn't surprise me that an automated tool is not perfect. The mitochondrial genome is small enough that manual verification can be feasibly done. You should at least look at the annotated information to make sure it makes sense; the response from NCBI suggests that it doesn't (e.g. nad5 extends into an annotated tRNA sequence).

As a direct / explicit response to your questions, just in case my answers are useful:

after applying MITOS, is there any manual or automatic review that should be applied?

I don't know. You should at least look at the annotated information to make sure it makes sense.

Is it common that the annotation returned by MITOS is not good enough?

I don't know; I have never used MITOS.

Or could that indicate that maybe there is a problem in the assembled genome? (looks unlikely except for a few SNPs as three assemblers return almost the same sequence)

I don't know; you haven't provided the assembled gene sequence or annotation. The response from NCBI suggests that there is a problem with either the sequence or annotation (e.g. nad5 extends into an annotated tRNA sequence).

Should we edit the annotation ourselves or there is a software that does it itself?

If there is an error in the annotation through automated methods, you should fix it manually.

How often using MITOS is enough or how often manually editing is needed?

I don't know; I have never used MITOS.

Answer 2

You might be luckier by submitting to INSDC by the ENA gate instead of NCBI.

To do so you first have to convert your GFF into the EMBL format used by ENA with the help of EMBLmyGFF3. During the conversion you can use the --tranlate option that can help to see if there is anything wrong in the ORFs (e.g. frameshifts).

Incomplete genes is supposed to be supported by INSDC databases, so I don't get why NCBI complains about it. Either their pipeline does not take care of this, or it is something specific to Mitochondrion annotations. Using EMBLmyGFF3 incomplete CDS will be handled properly and the location field in the resulting EMBL flat file will have the information that it is incomplete (see 3.4.3 Location examples in the documentation http://ftp.ebi.ac.uk/pub/databases/embl/doc//FT_current.html#3.4)

Then once you have your EMBL flat file you can check by yourself if there is anything not accepted by the DB using the ENA webin-cli (https://github.com/enasequence/webin-cli/releases/tag/v3.7.0) It there are errors they will be listed with details and you will have to fix your file accordingly. When everything is fixed you can submit to ENA. Once accepted by ENA it will be within 24 hours within NCBI.