Ñèáèðñêîå îòäåëåíèå ÐÀÍ 
Èíñòèòóò öèòîëîãèè è ãåíåòèêè



First Workshop on Information Technologies Application to Problems of Biodiversity and Dynamics of Ecosystems in North Eurasia (WITA-2001)

July 9-14, 2001, Novosibirsk, Russia

Abstracts


Genetic Bases of Biodiversity

New resource for prediction of subcellular location of proteins.

Bachinsky A., Ivanisenko V., Seledtzov I., Solovjev V.

SRC VB "Vector" (Koltsovo)

The decryption of sequences of genomes is only the first step in study of genetic systems.

One of the following steps, alongside with definition of possible functions of potential proteins, is prediction of subcellular and/or extracellular locations of these proteins.

Resources, known in the Internet making such predictions give many errors, therefore there was a problem on creation of a new system.

ProtComp is intended for the analysis of eucariotic sequences and combines in itself different methods of the prediction.

The tested sequence is compared to all proteins of the bank SWISS-PROT, for which the location is known. The matching is made by two modes: with the help of known algorithm BLASTA, and with usage of patterns of the specially constructed bank which has been carried out using PROF_PAT technology (http://wwwmgs.bionet.nsc.ru/mgs/programs/prof_pat/).

The system predicts signal and other sequences (signal-anchors, mitochondrial and chloroplast transit peptides, transmembrane segments) with usage of Ntural Nets approach. and also searches for motifs, characteristic for proteins of the defined localizations. Besides the charge of domains were between transmembrane segments (or, accordingly, N- and C-tails of a sequence) is defined, that allows to eelucidate membrane topology.

Neural nets differentiate the proteins on the following compartments: Cytoplasmic, Endoplasmic reticulum, Plasma membrane, Nuclear, Extracellular (Secreted), Mitochondrial, Chloroplast (for plants only), Peroxisomal, Golgi and Lysosomal (for animals and fungi only).

On the basis of outcomes of all these subsystems the integral score is formed which determines the prediction (Table 1).

ProtComp was tested on TrEMBL sequences, not including in training sets. The outcomes are represented in Table 2.

The resource is accessible to the anonymous users on the server http://www.softberry.com.

Table 1. An example of output of the ProtComp system for prediction of protein location.

Table 2. Results of testing ProtComp on TrEMBL sequences with known locations.

Note. Abstracts are published in author's edition


|Home Page| |English Part| [WITA2001]
Go to Home
© 1996-2000, Siberian Branch of Russian Academy of Sciences, Novosibirsk
    Last update: 06-Jul-2012 (11:44:54)