<div dir="ltr"><div class="gmail_extra"><div dir="ltr">Hello Ensembl Team,</div><div dir="ltr"><br></div><div>I recently downloaded the recently updated GFF3 release (MIPS v22). Thank you for fixing the problems I mentioned in my previous email!</div>
<div><br></div><div>However, there is still one major problem that is causing parsers to die, and a minor problem which will help the parsers. The minor suggestion is to start including the standard '###' strings on newlines for each logical feature block for a given seq_id in the first colunmn. This helps parsers know that its the end of that feature block and allows them to clear memory for the next section. Since your GFF3 does not have any of these strings on newlines, parsers have to keep everything in memory.</div>
<div><br></div><div>Second, there is scoping problems for the three_prime_UTR and five_prime_UTR definitions.</div><div><br></div><div>Changing this line:</div><div><font face="courier new, monospace">IWGSC_CSS_3AS_scaff_369935 . five_prime_UTR 199 200 . - . Parent=Traes_3AS_775C097A2.1</font><br>
</div><div><br></div><div>to:</div><div><font face="courier new, monospace">IWGSC_CSS_3AS_scaff_369935 . five_prime_UTR 199 200 . - . Parent=transcript:Traes_3AS_775C097A2.1<br></font></div>
<div><br></div><div>fixes the problem. After writing a py script to prefix 'transcript:' to the parent string ID for both five_prime and three_prime UTRs, I was able to successfully validate and load the GFF3s to standard visualization tools. If you would like to download the list of UTRs that have this problem, here is the an error log that you can inspect: <a href="http://169.237.215.34/ftp/temp/test_mips_gff3load.err">http://169.237.215.34/ftp/temp/test_mips_gff3load.err</a> (7.6MB) .<br>
</div><div><br></div><div>Cheers,</div><div>-Hans</div>
</div></div>