Synthesize exons when they are not included in the input GFF3

[garrettjstevens]

Sometimes we see a GFF3 that does not explicitly state where the exons are, e.g.

ctgA	example	gene	1050	9000	.	+	.	ID=EDEN;Name=EDEN;Note=protein kinase
ctgA	example	mRNA	1050	9000	.	+	.	ID=EDEN.1;Parent=EDEN;Name=EDEN.1;Note=Eden splice form 1;Index=1
ctgA	example	mRNA	1050	9000	.	+	.	ID=EDEN.2;Parent=EDEN;Name=EDEN.2;Note=Eden splice form 2;Index=1
ctgA	example	five_prime_UTR	1050	1200	.	+	.	Parent=EDEN.1
ctgA	example	five_prime_UTR	1050	1200	.	+	.	Parent=EDEN.2
ctgA	example	CDS	1201	1500	.	+	0	Parent=EDEN.1
ctgA	example	CDS	1201	1500	.	+	0	Parent=EDEN.2
ctgA	example	mRNA	1300	9000	.	+	.	ID=EDEN.3;Parent=EDEN;Name=EDEN.3;Note=Eden splice form 3;Index=1
ctgA	example	five_prime_UTR	1300	1500	.	+	.	Parent=EDEN.3
ctgA	example	CDS	3000	3902	.	+	0	Parent=EDEN.1
ctgA	example	five_prime_UTR	3000	3300	.	+	.	Parent=EDEN.3
ctgA	example	CDS	3301	3902	.	+	0	Parent=EDEN.3
ctgA	example	CDS	5000	5500	.	+	0	Parent=EDEN.1
ctgA	example	CDS	5000	5500	.	+	0	Parent=EDEN.2
ctgA	example	CDS	5000	5500	.	+	1	Parent=EDEN.3
ctgA	example	CDS	7000	7600	.	+	1	Parent=EDEN.3
ctgA	example	CDS	7000	7608	.	+	0	Parent=EDEN.1
ctgA	example	CDS	7000	7608	.	+	0	Parent=EDEN.2
ctgA	example	three_prime_UTR	7601	9000	.	+	.	Parent=EDEN.3
ctgA	example	three_prime_UTR	7609	9000	.	+	.	Parent=EDEN.1
ctgA	example	three_prime_UTR	7609	9000	.	+	.	Parent=EDEN.2

In this case we need to synthesize the exons for our internal representations.

We can use the five_prime_UTR, three_prime_UTR, and CDS lines to figure out where the exons are. If a UTR and a CDS are adjacent, they should be combined into a single exon. Otherwise, each unique CDS location should get an exon with the same location.

This needs to be handles in packages/apollo-shared/src/GFF3/gff3ToAnnotationFeature.ts. We'll probably want to check after processedCDS are determined in that file if there are any exons, and then synthesize them at that point if not.

[dariober]

This should be fixed in PR #492

The example in the issue looks like:

The code may be inefficient but the slow down doesn't seem noticeable. These are two replicates loading a small gff with 283 mRNAs:

# Before (just before this branch started):
git checkout 143354232ff73b042aa0f55996b6b94068eeb748 
time yarn dev feature import --profile testAdmin ~/Downloads/TGGT1_chrII.gff -d -a ToxoDB-67_TgondiiGT1_Genome.fasta.gz
progress [========================================] 100% | ETA: 0s | 605768/605768

real	0m20.265s
user	0m3.263s
sys	0m0.403s

time yarn dev feature import --profile testAdmin ~/Downloads/TGGT1_chrII.gff -d -a ToxoDB-67_TgondiiGT1_Genome.fasta.gz
progress [========================================] 100% | ETA: 0s | 605768/605768

real	0m18.901s
user	0m2.900s
sys	0m0.371s

Current:

git switch -
Previous HEAD position was 14335423 Make feature type ontology configurable (#472)
Switched to branch 'import_gff3_wo_exons_issue491'
Your branch is ahead of 'origin/import_gff3_wo_exons_issue491' by 2 commits.
  (use "git push" to publish your local commits)
time yarn dev feature import --profile testAdmin ~/Downloads/TGGT1_chrII.gff -d -a ToxoDB-67_TgondiiGT1_Genome.fasta.gz
progress [========================================] 100% | ETA: 0s | 605768/605768

real	0m19.276s
user	0m2.675s
sys	0m0.345s

time yarn dev feature import --profile testAdmin ~/Downloads/TGGT1_chrII.gff -d -a ToxoDB-67_TgondiiGT1_Genome.fasta.gz
progress [========================================] 100% | ETA: 0s | 605768/605768

real	0m19.263s
user	0m2.885s
sys	0m0.347s