Can metaeuk enforce splice site checking? #83
Description
Thank you for developing metaeuk! We are trying to supplement trinity assemblies with metaeuk-predicted "transcripts" for annotating a fish genome.
During the PASA step, we saw that among the ~64000 metaeuk-predicted transcripts, 23800 failed splice-site checks and were discarded by PASA.
I'm wondering if metaeuk checks for splice donor and acceptor sites when scanning for potential exons? Would adding that functionality improve accuracy somehow?
Here are some examples:
custom om_12116 12037 611 omscf_1 9 - ? ERROR 1405060-1416102 99.17 100 0 orient(a-/s?) align: 1405060(1926)-1405384(1602)>CT....AC<1408941(1601)-1409178(1363)>tt....tt<1413013(1362)-1413087(1288)>ca....tc<1413755(1287)-1413992(1050)>gg....ta<1414126(1049)-1414382(793)>GT....aa<1414453(792)-1414599(646)>tt....tt<1415298(645)-1415591(352)>aa....ga<1415669(351)-1415846(174)>tg....AG<1415941(173)-1416102(1) Splice site validations failed custom om_19938 19781 618 omscf_1 7 - ? ERROR 1412975-1416088 98.99 100 0 orient(a-/s?) align: 1412975(1386)-1413091(1270)>GC....AC<1413759(1269)-1413996(1032)>ga....AT<1414130(1031)-1414386(775)>ca....ta<1414457(774)-1414603(628)>ac....ca<1415302(627)-1415595(334)>ca....tt<1415673(333)-1415850(156)>tg....ct<1415945(155)-1416088(1) Splice site validations failed custom om_18083 17936 630 omscf_1 11 - ? ERROR 1418808-1423461 99 100 0 orient(a-/s?) align: 1418808(1485)-1418942(1351)>CT....AC<1419333(1350)-1419423(1260)>CT....AC<1419507(1259)-1419647(1119)>AT....ca<1420455(1118)-1420594(979)>tc....ca<1420976(978)-1421133(821)>ga....tt<1421666(820)-1421783(703)>GT....cc<1422214(702)-1422299(617)>cc....GC<1422467(616)-1422554(529)>ga....AT<1422632(528)-1422856(304)>ag....AC<1422996(303)-1423204(95)>CT....tt<1423374(94)-1423461(1) Splice site validations failed