Based on the data and questions you provided, here is a clear, step-by-step guide on how to proceed with the GenBank annotation for strains 2605 and 2914.

1. Minimum Contig Length for GenBank Submission

You do not strictly have to submit only contigs ≥ 500 bp, but it is highly recommended.

• Currently, GenBank’s technical minimum length for a Whole Genome Shotgun (WGS) sequence is 200 bp [[46]].
• However, contigs shorter than 500 bp rarely contain meaningful genes and are often assembly artifacts or sequencing noise. Filtering them out results in a cleaner, more biologically relevant submission.
• Future-proofing note: NCBI has announced that starting in January 2027, the minimum length requirement for prokaryotic and eukaryotic genome submissions will be raised to 1,000 nucleotides [[47]]. Adopting a ≥ 500 bp (or even ≥ 1,000 bp) cutoff now is a very safe and forward-looking strategy.

(Based on your data, filtering at ≥ 500 bp will leave you with 67 contigs for strain 2605 and 93 contigs for strain 2914, cleanly removing the tiny, high-depth noise contigs at the end of your lists).

2. Chromosomal vs. Plasmid Contigs (Depth Analysis)

Your assumption is absolutely correct.

• The contigs ranking 1–40 with a depth of ~1.0x (0.9x–1.1x) and larger lengths are almost certainly the main chromosomal fragments.
• The smaller contigs with significantly higher depth (e.g., 2x to >10x, such as contigs 48, 49, 51, 54 in strain 2605, and 46, 71, 83 in strain 2914) are classic signatures of plasmids, phages, or repetitive elements.
• It is excellent that you manually verified the circular=true contigs via BLAST. You should keep the circular=true flag in the FASTA headers for these, as the NCBI annotation pipeline explicitly recognizes this flag and will treat them appropriately.

3. Do You Need to Manually BLAST All Remaining Contigs?

No, manual web BLAST is not recommended. Checking dozens of contigs one by one via the NCBI web BLAST service is highly inefficient and unnecessary for a GenBank submission.

Instead, you should rely on the NCBI Prokaryotic Genome Annotation Pipeline (PGAP). PGAP is specifically designed to annotate draft WGS genomes consisting of multiple contigs, including both chromosomes and plasmids [[49]]. It will automatically:

• Identify and annotate plasmid-specific features (e.g., replication initiation proteins, plasmid partitioning proteins, mobility genes).
• Distinguish between chromosomal and plasmid sequences based on gene content and context.

If you or your co-author HK want to be absolutely certain about the nature of the remaining medium-length contigs before submission, you can run an automated local screen using tools like PlasmidFinder or a batch local BLAST against a plasmid database, rather than doing it manually via the web interface.

4. Recommended Actionable Workflow

To efficiently fulfill HK’s request, follow these steps:

1. Filter the FASTA files: Write a simple script (e.g., using awk or Biopython) to remove all contigs < 500 bp from both the 2605 and 2914 assemblies. Ensure the circular=true flag remains in the defline of the confirmed plasmids.
2. Prepare Metadata: Ensure you have the required BioProject and BioSample accession numbers, along with the strain names, isolation sources, and assembly method details ready.
3. Submit via NCBI Submission Portal: Use the NCBI Submission Portal for Prokaryotic Genomes [[54]]. Upload your filtered FASTA files and select the option to have NCBI annotate the genomes using PGAP.
4. Review the Output: Once PGAP processing is complete, you can download the annotated GenBank (.gbff) files. You can then easily search these files for “plasmid” or review the locus tags to confirm which contigs were annotated as plasmids, and share this comprehensive annotation report with HK.

This approach minimizes manual labor, adheres to GenBank best practices, and leverages NCBI’s robust automated pipeline to give you high-quality annotations. Let me know if you need help writing the script to filter the FASTA files!