First NameSteven
Last NameJohnson B.S .
Emailjohnsonsd@ukzn.ac.za
Supervisor Name Dr . SeanEddy
UniversityWASHINGTON UNIVERSITY
CountryUnited States
Keywordssequence comparison, algorithms introduce, protein knowledge, Methods
Publication Date21 July, 2015
DegreePhD
DomainBiology

REMOTE PROTEIN HOMOLO GYDETE CTIONUSING HIDDEN MARKOV MODELS

Abstract

Th e fi r st sequ en ce com p ar ison algor it h m s wer e int r o d u ced over 30 year s ago. Th e m ot ivat ion b eh in d t h at wor k st ill exist s t o d ay, t o con n ect t h e vast r eser voir of exist in g p r ot ein kn owled ge t o lesser kn own sequ en ces. If two p r otein s ar e d eem ed h om ologou s, t h en in for m at ion r egar d in g fu n ct ion an d st r u ct ur e m ay b e com m on b etween t h em . At t h e least , it p r ovid es a b asis for hyp ot h eses and exp er im ent at ion . S equ en ce com p ar ison m et h o d s, an d t h e d at a access t h at t h ey r ely on , h ave gr own con sid er ab ly over t h e d ecad es. Th is gr owt h h as p r im ar ily com e fr om a con st ant cycle of in cor p or at in g b iological sou r ces of in for m at ion int o sear ch algor it h m s, t h e id ent ifi cat ion of m or e d ist ant h om ologies, an d t h e fu r t h er r efi n em ent of in for m at ion gain ed fr om t h e n ew h om ologies into t h e n ext gen er at ion of algor it h m s an d sear ch st r at egies. Th e fo cu s of t his r esear ch h as b een in t h e u se of Hid d en M ar kov M o d els ( HM M s) for r em ot e p r ot ein hom ology d et ect ion . We h ave im p lem ent ed in for m at ion cont ent sequ en ce weight in g t o im p r ove lo cal align m ent sear ch es. We id ent ifi ed t h at t h e For war d nu ll scor e d ist r ib u t ion can b e ap p r oxim at ed by t h e exp on ent ial d ist r ib ut ion an d u t ilized t h is t o calcu lat e accu r at e an d r ob u st P – valu es for HHM For war d scor es. In ad d it ion , we h ave d evelop ed an HM M sear ch h eu r ist ic t o d ecr ease t h e t im e r equ ir ed t o sear ch lar ge d at ab ases. Fu r t h er , we h ave in cor p or at ed t h is h eur ist ic int o an it er at ive HM M sear ch m et h o d t h at sign ifi cant ly in cr eases sear ch sen sit ivity.

Download Thesis