Using HHBLITS software improves the method of comparing protein chains

A group of Bioinformatics leading by Dr. Johannes Soding, working at the Center for Genetics, Ludwig Maximilian University Munich, Germany, has developed a new search method (using HHblits software) to compare sequences of proteins with similar sequences in public databases: faster and more than twice as much as protein-related evolutionary signs compared to previous methods; list the properties of proteins (through functional and structural analysis of proteins).

This new search method (using HHblits software ) will be used in analyzing the structure and function of many proteins.

Proteins participate in almost all biochemical processes of life. The functions that a protein performs largely depend on the sequence of building blocks of 20 amino acids (amino acids) and the three-dimensional structure of pleated amino acids. From the similarities of protein chains, bioinformatics methods can predict the evolutionary relationship of proteins, with similar structures and functions.

This means that the sequence of proteins that need to be studied will be compared to millions of protein sequences (in pairs) in public databases (with pre-engineered structures and functions). ). The general relationships between sequences and functions that enable HHblits software to predict the structure and function of a certain protein by comparing its sequence with proteins (with the same structure and function) Features) are available in Public Database.

HHblits software, superior to PSI-BLAST software (which is the most popular tool in the last 15 years, is used for comparing protein sequences because: high speed, sensitivity and accuracy ) in all aspects of performance. The researchers first converted the sequence of proteins to be analyzed and sequenced in public databases (based on the algorithm of finding hidden Markov Models (HMMs) ). The HMMs algorithm is a statistical model that combines mutation probabilities defined from sequencing, so this step increases the sensitivity and accuracy of subsequent similar searches.

In addition, the team developed a filtering process, which allows a significant reduction in the amount of data that needs to be searched. The trick is before assembling similar protein sequences from public databases. Each linked column is then labeled with one of the 219 "letters", (ie columns with the same amino acid composition (amino acid) will be represented by the same character)."By translating protein sequences arranged in a sequence of 219 characters, we can alternately compare protein pairs. This helps to reduce the search time by 2500 times . " Besides, the researchers also combined information about the three-dimensional structures of proteins.