How does ncbi-blastP calculate the score ? (Why it is lower than mean-diagonal(blossum62) * length ) ?

Question

[Alexander]

Question briefly: Let us take an example of SRC human protein (P12931) - length 536 - on top of the blastP output we will see SRC itself - with score 1119 . Why ?

What bothers me - diagonal of the blossum62 matrix - 4,5,6,9.. so in my mind output should be greater than at least 4*536 . But it is not the case. I thought the score is calculated as here: ChatGPT optimized for bioinformatics questions , i.e. in the case of the complete coincidence we should sum up over symbols blossum62[symbol, symbol], is not it ?

Note that BioPython (or Parsail) gives score 2834 , Dimond gives 1085 (almost like 536*2 = 1072, why ? ) (Example here: https://www.kaggle.com/code/alexandervc/protein-aligners-benchmark-parasail-diamond-etc?scriptVersionId=132991490&cellId=47 )

More details: SRC human protein (P12931) - uniprot:

https://www.uniprot.org/uniprotkb/P12931/entry sequence:

MGSNKSKPKDASQRRRSLEPAENVHGAGGGAFPASQTPSKPASADGHRGPSAAFAPAAAEPKLFGGFNSSDTVTSPQRAGPLAGGVTTFVALYDYESRTETDLSFKKGERLQIVNNTEGDWWLAHSLSTGQTGYIPSNYVAPSDSIQAEEWYFGKITRRESERLLLNAENPRGTFLVRESETTKGAYCLSVSDFDNAKGLNVKHYKIRKLDSGGFYITSRTQFNSLQQLVAYYSKHADGLCHRLTTVCPTSKPQTQGLAKDAWEIPRESLRLEVKLGQGCFGEVWMGTWNGTTRVAIKTLKPGTMSPEAFLQEAQVMKKLRHEKLVQLYAVVSEEPIYIVTEYMSKGSLLDFLKGETGKYLRLPQLVDMAAQIASGMAYVERMNYVHRDLRAANILVGENLVCKVADFGLARLIEDNEYTARQGAKFPIKWTAPEAALYGRFTIKSDVWSFGILLTELTTKGRVPYPGMVNREVLDQVERGYRMPCPPECPESLHDLMCQCWRKEPEERPTFEYLQAFLEDYFTSTEPQYQPGENL

BlastP: https://blast.ncbi.nlm.nih.gov/Blast.cgi?PROGRAM=blastp&PAGE_TYPE=BlastSearch&LINK_LOC=blasthome

Answer 1

[shelkmike]

The scores that you see are "bit scores". For how they are calculated see https://www.ncbi.nlm.nih.gov/BLAST/tutorial/Altschul-1.html#head3

[Alexander]

Thanks for the answer ! Would be happy to see you in our telegram group: @sberlogabio