Python Programming for Biology: Bioinformatics and Beyond

Sequence classification

You  will  often  find  this  task  referred  to  as  sequence  annotation,  i.e.  to  associate  extra

information with a sequence to say what it is or does, or perhaps at least say what it might

do.  The  principle  behind  this  is  to  compare  a  query  sequence  (maybe  newly  discovered)

with  other  sequences  that  we  have  some  information  about.  If  the  query  can  be  aligned

with  some  of  these  other  sequences  then  we  can  often  infer  that  our  query  sequence  has

some  of  the  same  properties  as  those  that  it  matches,  and  the  degree  of  similarity  says

something  about  the  confidence  of  the  inference.  This  can  be  done  with  all  the  different

kinds  of  biological  polymer  molecule:  DNA,  RNA  and  protein.  An  illustrative  example

would  be  when  the  elephant  genome  sequence  is  being  processed,  because  there  is  a

massive amount of information about the genomes of other mammals (human, mouse, rat,

cat, dog etc. …) then there’s a good chance that any given gene will align well with other

genes that we already have knowledge of. Accordingly, if we find the elephant gene that

aligns  best  with  mammal  genes  labelled  as  beta-globin  then  it  is  highly  likely  that  this

gene is the elephant’s version of beta-globin and that it functions in the same way, making

up the haemoglobin protein to carry oxygen in red blood cells. On this occasion we could

align  either  DNA  sequences  or  protein  sequences,  as  we  are  working  with  relatively

closely related species. In situations where the degree of similarity is not so good protein

sequence alignments can yield better results than DNA; because, for reasons that will be

discussed  later,  protein  sequence  similarity  tends  to  be  better  preserved  during  evolution

than for DNA.