updated README for submission

08-submission/README.md

 # Submission
 
-Utility scripts for data upload to ENA.
-
-- [EMBL format conversion](convert_to_embl.sh) for an annotated genome (Fasta + GFF3) - inspired
-  from a [prokka issue](https://github.com/tseemann/prokka/issues/145), leading to
-  [GFF3toEMBL](https://github.com/sanger-pathogens/gff3toembl) and eventually to the real solution,
-  [EMBLmyGFF3](https://github.com/NBISweden/EMBLmyGFF3).
-- A [Python script](txome_manifest.py) to batch-create manifest files for de novo transcriptome
-  submissions to ENA.
\ No newline at end of file
+Utility scripts for data upload to ENA. This was very difficult to figure out.
+
+## Raw data submission
+
+Constructing manifest files for the raw data was fairly straightforward following the
+[ENA documentation](https://ena-docs.readthedocs.io/en/latest/index.html), especially the part
+concerning [raw reads submission](https://ena-docs.readthedocs.io/en/latest/submit/reads.html).
+
+## Assembled transcriptome submission
+
+Since we had a lot of time points, we wrote a custom [Python script](txome_manifest.py) to
+batch-create manifest files for all the _de novo_ assembled transcriptomes. This was also relatively
+straightforward, especially the transcriptome assembly [submission
+guide](https://ena-docs.readthedocs.io/en/latest/submit/assembly/transcriptome.html).
+
+## Assembled genome with functional annotation
+
+This was a pain to get right, and required two-three weeks to get right; and we still weren't able
+to include functional information like predicted functions, PFAM domains, or EC numbers.
+
+ENA requires that annotated genome assemblies be submitted as EMBL files. To control the process, we
+wrote a [Makefile](./Makefile). This was an excellent decision to make early in the submission
+procedure, as we ended up having to finetune and re-run the entire thing multiple times. The steps
+are:
+
+1. [compose](gff-01-compose_gff.sh) the GFF by combining the protein coding gene models (Isoseq,
+   BRAKER3 round 1, BRAKER3 round 2, de novo assembled transcriptomes), and the tRNA prediction. We
+   also manually replaced three gene models that were incorrectly split up with an earlier BRAKER3
+   gene model. At this stage, we officially renamed pseudochromosomes 54-59 to 52-57, and sort the 
+   resulting GFF file.
+2. [annotate](gff-02-functional_annot.ipynb) the GFF by adding information from EggNOG-mapper. We
+   need a [wrapper script](gff-02-functional_annot.sh) for that, since using conda environments is
+   rather tricky with `make`. The information added, if present, is the proposed gene symbol,
+   function description, EC number, and PFAM domains identified.
+3. [conform](gff-03-ENA_conform.sh) the GFF to ENA standards. This step flags short introns (less
+   than 10 bases, presumably created by polymerase slippage or post-transcriptional modifications)
+   with the `pseudo` tag, ensuring that the ENA validator doesn't complain about them. A previous
+   version (commented out) [extracted](gff-03-build_kill_list.py) the offending exons into a new GFF
+   file; this approach was abandoned because it had to extract entire genes, which was beside the
+   point.
+4. [convert](gff-04-convert_to_embl.sh) the GFF to EMBL format by combining it with the FASTA file.
+   Uses [EMBLmyGFF3](https://github.com/NBISweden/EMBLmyGFF3) (and a lot of the useful information
+   provided in its solved and unsolved GitHub issues, the true source of most useful information
+   about ENA). Before running the real script, one has to run `EMBLmyGFF3` with the
+   `--expose-translations` and modify the resulting .json file in order to suppress exons (as this
+   will create the "abutting features cannot be adjacent" error that the validator throws).
+5. [submit](gff-05-submit_to_ENA.sh) the EMBL file to ENA. The manifest file and the chromosome file
+   still have to be written separately.
+
+Big thanks to [Jacques Dainat](https://github.com/Juke34), who wrote and is maintaining
+[AGAT](https://github.com/NBISweden/AGAT) and [EMBLmyGFF3](https://github.com/NBISweden/EMBLmyGFF3),
+as well as diligently answering questions and resolving issues.
\ No newline at end of file