Hi Elizabeth, If you encode the transposon as match/match_part, they'll show up in the results panel (the black panel). To enconde the grouping, you'll want to do something like this:
<gff3> ##gff-version 3 MRGH6_contig ltr_finder:def transposable_element 55509 64198 6 - . Name=transposable_element_1;ID=transposable_element_1 MRGH6_contig ltr_finder:def three_prime_LTR 62941 64198 6 - . Name=three_prime_LTR;Source=ltr_finder:def;Strand=-;ID=three_prime_LTR_1;Parent=transposable_element_1 MRGH6_contig ltr_finder:def primer_binding_site 62841 62857 6 - . Name=primer_binding_site;Source=ltr_finder:def;Strand=-;ID=primer_binding_site_1;Parent=transposable_element_1;Parent=transposable_element_1 MRGH6_contig ltr_finder:def RR_tract 56801 56815 6 - . Name=RR_tract;Source=ltr_finder:def;Strand=-;ID=RR_tract_1;Parent=transposable_element_1 MRGH6_contig ltr_finder:def five_prime_LTR 55509 56766 6 - . Name=five_prime_LTR;Source=ltr_finder:def;Strand=-;ID=five_prime_LTR_1;Parent=transposable_element_1 </gff3> Note that every feature has a different ID and the child features link back to the transposon using the "Parent" attribute. If you add "column : ID" to your tiers file for the appropriate type, you'll get the ID added to the list of attributes so that you can figure out what each child represents. Cheers, Ed On Wed, 12 May 2010, elizabeth henaff wrote:
Hi Ed, I'm not quite sure how to define the groupings if I convert the types to match or match_part. It seems that if I use match_part it groups the features by the source (2nd) column. And I also tried assigning the same ID tag (in the 9th column) to each set of features that I'd like to be connected, but in that case they are still grouped by source, with only the first element that has a given ID showing up. Also, where is the type definition for match? I didn't find it in the gff3.tiers file. I've also tried renaming the types as transcript for the retrotransposons, and exons for the transposon parts, while preserving the original SO types in the Name tag. This seems to fit my needs in that I can use the Parent / Child relationships, without losing the information as to the SO type. If I can't find a way to display them as discontinuous features in the results section, I might stick with that work-around. Thanks for any ideas! Elizabeth On Wed, May 12, 2010 at 8:29 PM, Ed Lee <[email protected]> wrote: Hi Elizabeth, Are you trying to view these features in the annotation or results panel? For annotations, Apollo doesn't support multi-tier features other than gene-transcript-exon (and its variants). With the current implementation, you will not be able to group the children features together (and keep their correct SO type). If you want to view them in the results panel, you can just convert the types to match/match_part and you can get boxes connected by lines for grouping the individual features together. Cheers, Ed On Wed, 12 May 2010, elizabeth henaff wrote: Hi Ed, Thanks for your response. The data I have is such that I had one Parent (retrotransposon) with multiple children features (LTR, primer binding site, etc). It is meaningful for me to retain these children features, so I was wondering of it would be possible to circumvent the child-parent relationship by eliminating the Parent feature, and representing all its children as one discontinuous feature. I've tried to do that by adding an identical ID attribute to the children that belong together and "groupby : GENE" to the [type] definition but when I load the gff3 I only see the first feature with a given ID. this is what I have added to gff3.tiers: [Tier] tiername : LTR_de_novo expanded : true labeled : true curated : true [Type] label : LTR_retrotransposon tiername : LTR_de_novo datatype : primer_binding_site datatype : RR_tract datatype : five_prime_LTR datatype : three_prime_LTR datatype : target_site_duplication datatype : transposon_fragment glyph : ThinRectangle color : red utr_color : 250,250,210 column : GENOMIC_LENGTH 1 column : GENOMIC_RANGE 1 column : SCORE column : NAME column : Source column : Strand and here's a snippet of the gff3 file: ##gff-version 3 MRGH6_contig ltr_finder:def three_prime_LTR 62941 64198 6 - . Name=three_prime_LTR; Source=ltr_finder:def; Strand=-; ID=ltr_finder:def_ltr_finder_1 MRGH6_contig ltr_finder:def primer_binding_site 62841 62857 6 - . Name=primer_binding_site; Source=ltr_finder:def; Strand=-; ID=ltr_finder:def_ltr_finder_1 MRGH6_contig ltr_finder:def RR_tract 56801 56815 6 - . Name=RR_tract; Source=ltr_finder:def; Strand=-; ID=ltr_finder:def_ltr_finder_1 MRGH6_contig ltr_finder:def five_prime_LTR 55509 56766 6 - . Name=five_prime_LTR; Source=ltr_finder:def; Strand=-; ID=ltr_finder:def_ltr_finder_1 MRGH6_contig ltr_finder:def primer_binding_site 105081 105103 6 - . Name=primer_binding_site; Source=ltr_finder:def; Strand=-; ID=ltr_finder:def_ltr_finder_2 MRGH6_contig ltr_finder:def RR_tract 95013 95027 6 - . Name=RR_tract; Source=ltr_finder:def; Strand=-; ID=ltr_finder:def_ltr_finder_2 MRGH6_contig ltr_finder:def five_prime_LTR 94168 95008 6 - . Name=five_prime_LTR; Source=ltr_finder:def; Strand=-; ID=ltr_finder:def_ltr_finder_2 MRGH6_contig ltr_finder:def three_prime_LTR 105170 106020 6 - . Name=three_prime_LTR; Source=ltr_finder:def; Strand=-; ID=ltr_finder:def_ltr_finder_2 Thanks for your help! Elizabeth On Wed, May 12, 2010 at 6:15 PM, Ed Lee <[email protected]> wrote: Hi Elizabeth, The only parent-child relationships that Apollo currently supports is the gene-transcript-exon relationship. Unless you need to retain the LTR relationship, I would suggest choosing the feature that reflects the analysis best (either the parent or child) and then removing the other (if removing the child, remember to remove the "Parent" attribute in the GFF3 file). Cheers, Ed On Wed, 12 May 2010, elizabeth henaff wrote: Hello all, I'm trying to use Apollo to annotate transposable elements in some BAC sequences. I have run various prediction programs that generate data in GFF3 format, and my idea is to load these results into Apollo to be able to manually curate them. The nature of the data is such that there are Parent-Child relationships between features (as defined by the Parent=xxx tag in the 9th column), for example a three_prime_LTR feature would have as a Parent a LTR_retrotransposon feature, indicating that that LTR sequence belongs to a particular retrotransposon. When I load the gff3 file into Apollo, I get the following error message: "only transcripts and exons can be children of annotations: three_prime_LTR[three_prime_LTR]" and a similar error message for each feature that has a Parent tag. Apollo then starts up, but only the features that do not have a Parent tag appear. In gff3.tiers I have defined a new [tier] as TE_annotation and a [type] for each of the datatypes (the second column of the gff3 file) present in my gff3 file. For each of the [type] fields I defined the number_of_levels field as 3, but that doesn't seem to fix the problem. Any idea how I can resolve this? Thanks so much for any help! I've tried to attach the gff3, fasta and gff3.tiers files I'm using but it seems like it makes the message too big, but I'd be happy to send them if that would help. I'm running version 1.11.2 on ubuntu. Cheers, Elizabeth Centre de Recerca en AgriGenomica Barcelona, Espanya
_______________________________________________ apollo mailing list [email protected] http://mail.fruitfly.org/mailman/listinfo/apollo
