aragorn_out_to_gff3.py: Handle introns

[sjackman]

Does aragorn_out_to_gff3.py handle introns? It doesn't appear to after scanning the code. Feature request? Here's an example of a few tRNA with introns.

3   tRNA-Val              c[883351,883869]  35      (tac)i(40,443)
5   tRNA-seC               [244644,244785]  87      (tca)i(29,57)
7   tRNA-Arg               [615338,616255]  36      (tcg)i(40,822)
2   tRNA-Ser               [324667,326116]  35      (gga)i(40,1354)
1   tRNA-Val                c[84563,87002]  2383    (tac)i(34,2347)
1   tRNA-Asp              c[114384,114687]  39      (atc)i(43,206)
4   tRNA-seC               [324290,324369]  36      (tca)i(38,6)
5   tRNA-Ser               [345441,345776]  28      (gct)i(33,251)
7   tRNA-Val               [405066,405293]  34      (cac)i(35,133)
3   tRNA-Asn               [367853,369971]  2062    (att)i(32,2028)
3   tRNA-Pro               [234486,236430]  1887    (agg)i(31,1856)
1   tRNA-Ser                 [66020,66741]  32      (gct)i(37,633)
2   tRNA-Leu                 [84137,84224]  36      (tag)i(38,10)
3   tRNA-Arg               [168865,170209]  1285    (gcg)i(32,1253)
4   tRNA-His               [175766,176602]  776     (gtg)i(34,742)
1   tRNA-seC                c[88201,88280]  36      (tca)i(38,6)

[hexylena]

@sjackman it sure doesn't! I'd never seen results like that, but I only work with small viruses. How should this be marked up in the GFF3?

[hexylena]

Also, I don't suppose you could link me to a fasta sequence which would produce the data with introns, so I can test?

[sjackman]

Dang, you're fast. I'll get you some data.

[sjackman]

Here you go. Thanks! https://gist.github.com/sjackman/31cd4e17347ff1af488c

[hexylena]

@sjackman (never had comments on gists, I'm assuming it emails you, but in case it doesn't...)

would you mind sharing the command you ran this with? As of Aragorn v1.2.36 I didn't seem to be able to produce any headers with a )i( sequence in them, despite trying a number of configurations.

[sjackman]

No, I didn't get a notification from the gist. That seems like a GitHub issue/bug.

[sjackman]

Sorry, I meant to post the shell command and forgot.

aragorn -gcbact -i -c -w -o aragorn.tsv sample.fa

[hexylena]

brilliant, cheers. I'll ping you when I have a fix in place

[sjackman]

I'll mention that I've only ever seen a single intron. I have no idea whether ARAGORN could possibly output multiple introns. Thanks, Eric!

[hexylena]

Good to know. Having never worked with anything other than bacteriophages, I appreciate the insight.

[hexylena]

@sjackman not sure how familiar you are with the aragorn/maker/etc, but in the gff3 file you posted I'm having a bit of trouble rationalising a couple of the feature locations:

1   tRNA-Lys                  c[1533,4118]  34      (ttt)i(39,2512)
7   tRNA-Met                c[26588,26660]  34      (cat)
8   tRNA-Tyr                 [26850,27468]  559     (ata)i(36,523)

becomes:

1   maker   tRNA    1534    4115    100 -   .   ID=tRNA1;Parent=gene1;Name=trnK-UUU;_AED=0.48;_eAED=0.48;_QI=12|1|1|1|0|0|2|1|19
1   maker   tRNA    26588   26660   100 -   .   ID=tRNA7;Parent=gene7;Name=trnM-CAU;_AED=0.00;_eAED=0.00;_QI=0|-1|1|1|-1|0|1|1|24
1   maker   tRNA    26850   27468   100 +   .   ID=tRNA8;Parent=gene8;Name=trnV-UAC;_AED=0.48;_eAED=0.48;_QI=0|1|1|1|0|0|2|1|25

In the tRNA1, the location is modified, +1 on the left hand, and -3 on the right, whilst in tRNA2, the location is untouched. Given that the tRNA runs the full length of the feature, and that the other intron containing tRNA has an umodified location, I can't attribute it to the presence/absence of an intron. Is this a maker bug? Are you familiar enough to say?

I was just checking my work against the "reference", when I noticed a couple of these sorts of discrepancies.

[sjackman]

Hi, Eric. Very sorry. I should have mentioned that the GFF is generated not de novo but by aligning tRNA sequences from a difference (very closely related) species to the reference by MAKER using Exonerate. So, the coordinates of the de novo ARAGORN features and the MAKER/Exonerate features may not agree exactly.

[hexylena]

@sjackman no worries, not a problem. That explains the differences!

[bgruening]

Seems that I missed a lot! Thanks @erasche for fixing this.