The Blosum scoring matrices Morten Nielsen BioSys, DTU

Outline Alignment scoring matrices –What is a BLOSUM50 matrix and how is it different from a BLOSUM80 matrix? What are Blosum matrices good for? –Sequence alignment –Infer properties from one protein to another How can PSSM’s be used to improve the accuracy of a scoring matrix?

Sequence alignment SEQ1: VLSPADKTNVKAAWGKVGAHAGEYGAEALERMFLSFPTTKTYFPHFDLSHGSAQVKGHGKK VADALTNAVAHVDDPNALSALSDLHAHKLRVDPVNFKLLSHCLLVTLAAHLPAEFTPAVHA SLDKFLASVSTVLTSKYR SEQ2: VHLTPEEKSAVTALWGKVNVDEVGGEALGRLLVVYPWTQRFFESFGDLSTPDAVMGNPKVK AHGKKVLGAFSDGLAHLDNLKGTFATLSELHCDKLHVDPENFRLLGNVLVCVLAHHFGKEF TPPVQAAYQKVVAGVANALAHKYH Are these two proteins alike?

Amino acid properties Serine (S) and Threonine (T) have similar physicochemical properties Aspartic acid (D) and Glutamic acid (E) have similar properties Substitution of S/T or E/D occurs relatively often during evolution => Substitution of S/T or E/D should result in scores that are only moderately lower than identities =>

Protein substitution matrices A 5 R -2 7 N D C Q E G H I L K M F P S T W Y V A R N D C Q E G H I L K M F P S T W Y V BLOSUM50 matrix: Positive scores on diagonal (identities)‏ Similar residues get higher (positive) scores Dissimilar residues get smaller (negative) scores

Sequence Alignment 1PLC._ 1PLB._

Where is the active site? Sequence alignment 1K7C.A TVYLAGDSTMAKNGGGSGTNGWGEYLASYLSATVVNDAVAGRSARSYTREGRFENIADVVTAGDYVIVEFGHNDGGSLSTDN S G N 1WAB._ EVVFIGDSLVQLMHQCE---IWRELFS---PLHALNFGIGGDSTQHVLW--RLENGELEHIRPKIVVVWVGTNNHG K7C.A GRTDCSGTGAEVCYSVYDGVNETILTFPAYLENAAKLFTAK--GAKVILSSQTPNNPWETGTFVNSPTRFVEYAEL-AAEVA 1WAB._ HTAEQVTGGIKAIVQLVNERQPQARVVVLGLLPRGQ-HPNPLREKNRRVNELVRAALAGHP 1K7C.A GVEYVDHWSYVDSIYETLGNATVNSYFPIDHTHTSPAGAEVVAEAFLKAVVCTGTSL H 1WAB._ RAHFLDADPG---FVHSDG--TISHHDMYDYLHLSRLGYTPVCRALHSLLLRL---L

Homology modeling and the human genome

BLOSUM = BLOck SUbstitution Matrices Focus on conserved domains, MSA's (multiple sequence alignment) are ungapped blocks. Compute pairwise amino acid alignment counts –Count amino acid replacement frequencies directly from columns in blocks –Sample bias: Cluster sequences that are x% similar. Do not count amino acid pairs within a cluster. Do count amino acid pairs across clusters, treating clusters as an "average sequence". Normalize by the number of sequences in the cluster. BLOSUM x matrices –Sequences that are x% similar were clustered during the construction of the matrix.

Log-odds scores BLOSUM is a log-likelihood matrix: Likelihood of observing j given you have i is –P(j|i) = P ij /P i The prior likelihood of observing j is –Q j The log-likelihood score is –S ij = 2log 2 ((P(j|i)/Q j ) = 2log 2 (P ij /(Q i Q j ))

So what does this mean? An example N AA = 14 N AD = 5 N AV = 5 N DA = 5 N DD = 8 N DV = 2 N VA = 5 N VD = 2 N VV = 2 1: VVAD 2: AAAD 3: DVAD 4: DAAA P AA = 14/48 P AD = 5/48 P AV = 5/48 P DA = 5/48 P DD = 8/48 P DV = 2/48 P VA = 5/48 P VD = 2/48 P VV = 2/48 Q A = 8/16 Q D = 5/16 Q V = 3/16 MSA

So what does this mean? Q A =0.50 Q D =0.31 Q V =0.19 Q A Q A = 0.25 Q A Q D = 0.16 Q A Q V = 0.09 Q D Q A = 0.16 Q D Q D = 0.10 Q D Q V = 0.06 Q V Q A = 0.09 Q V Q D = 0.06 Q V Q V = 0.03 P AA = 0.29 P AD = 0.10 P AV = 0.10 P DA = 0.10 P DD = 0.17 P DV = 0.04 P VA = 0.10 P VD = 0.04 P VV = : VVAD 2: AAAD 3: DVAD 4: DAAA MSA

So what does this mean? Q A Q A = 0.25 Q A Q D = 0.16 Q A Q V = 0.09 Q D Q A = 0.16 Q D Q D = 0.10 Q D Q V = 0.06 Q V Q A = 0.09 Q V Q D = 0.06 Q V Q V = 0.03 P AA = 0.29 P AD = 0.10 P AV = 0.10 P DA = 0.10 P DD = 0.17 P DV = 0.04 P VA = 0.10 P VD = 0.04 P VV = 0.04 BLOSUM is a log-likelihood matrix: S ij = 2log 2 (P ij /(Q i Q j )) S AA = 0.44 S AD =-1.17 S AV = 0.30 S DA =-1.17 S DD = 1.54 S DV =-0.98 S VA = 0.30 S VD =-0.98 S VV = 0.49

The Scoring matrix ADV A D V : VVAD 2: AAAD 3: DVAD 4: DAAA MSA

And what does the BLOSUMXX mean? Cluster sequence Blocks at XX% identity To statistics only across clusters Normalize statistics according to cluster size A) B) AV AP AL VL AV GL GV } min XX % identify

And what does the BLOSUMXX mean? A) B) AV AP AL VL AV GL GV

And what does the BLOSUMXX mean? High Blosum values mean high similarity between clusters –Conserved substitution allowed Low Blosum values mean low similarity between clusters –Less conserved substitutions allowed

BLOSUM80 A R N D C Q E G H I L K M F P S T W Y V A R N D C Q E G H I L K M F P S T W Y V = 9.4 = -2.9

BLOSUM30 A R N D C Q E G H I L K M F P S T W Y V A R N D C Q E G H I L K M F P S T W Y V = 8.3 = -1.16

How do Blosum scoring matrices link to Blosum frequences matrices? The blosum frequency matrix is –P(i|j) The Blosum scoring matrix is So there is a one-to-one mapping between the two matrices