Protein Sequence Prediction Research Articles

EC number prediction of protein sequences based on combination of hierarchical and global features.

The identification of enzyme functions plays a crucial role in understanding the mechanisms of biological activities and advancing the development of life sciences. However, existing enzyme EC number prediction methods did not fully utilize protein sequence information and still had shortcomings in identification accuracy. To address this issue, we proposed an EC number prediction network using hierarchical features and global features (ECPN-HFGF). This method first utilized residual networks to extract generic features from protein sequences, and then employed hierarchical feature extraction modules and global feature extraction modules to further extract hierarchical and global features of protein sequences. Subsequently, the prediction results of both feature types were combined, and a multitask learning framework was utilized to achieve accurate prediction of enzyme EC numbers. Experimental results indicated that the ECPN-HFGF method performed best in the task of predicting EC numbers for protein sequences, achieving macro F1 and micro F1 scores of 95.5% and 99.0%, respectively. The ECPN-HFGF method effectively combined hierarchical and global features of protein sequences, allowing for rapid and accurate EC number prediction. Compared to current commonly used methods, this method offers significantly higher prediction accuracy, providing an efficient approach for the advancement of enzymology research and enzyme engineering applications.

Abstract 7418: LongFuse: Detecting gene fusion transcripts from high throughput long-read single cell RNA sequencing data

Abstract Gene fusion resulted from genomic translocation plays important oncogenic roles including but not limited to gene activation and changing protein coding sequences. Historically, fusion gene analysis has been limited to bulk levels resulting in mixed oncogene activation models. Single cell Nanopore RNA sequencing (scNanoRNAseq) has the unique strength in detecting full-length chimeric transcripts in thousands of single cells in parallel. However, robust computational tools for detecting fusion genes in single cells from scNanoRNAseq data are missing. Here, we developed LongFuse, a fast and accurate tool, to detect fusion genes and their splicing isoforms in single cells from scNanoRNAseq data. LongFuse implements an XOR logic gate algorithm to speed up fusion candidate searching and stringent criteria to eliminate false discoveries due to trans-splicing events and randomly ligated chimeras. We applied LongFUSE onto scNanoRNAseq data of both human prostate cancer cell line, VCaP, and tumor tissues with known TMPRSS2-ERG fusion. Our results showed that both cancer cell line and prostate tumors harbored a mixture of fusion positive and fusion negative tumor cells. Gene expression analysis revealed transcriptional differences between fusion positive and negative cells including genes involved in cancer cell proliferation and invasion programs such as ELL2, ZBTB16, IGF1R, ANK3, MYPOP and PLPP1 genes. LongFUSE also calculates the splicing isoforms of fusion gene transcripts, allowing prediction of protein sequences of fusion gene transcripts. Taken together, our data demonstrated that LongFUSE is a robust computational tool for detecting gene fusions and their splicing isoforms at single cell level from scNanoRNAseq data. Citation Format: Cheng-Kai Shiau, Yueying He, Hsiao-Yun Lin, Lina Lu, Xiao-Dong Lu, Jindan Yu, Ruli Gao. LongFuse: Detecting gene fusion transcripts from high throughput long-read single cell RNA sequencing data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7418.

MLapRVFL: Protein sequence prediction based on Multi-Laplacian Regularized Random Vector Functional Link

Protein sequence classification is a crucial research field in bioinformatics, playing a vital role in facilitating functional annotation, structure prediction, and gaining a deeper understanding of protein function and interactions. With the rapid development of high-throughput sequencing technologies, a vast amount of unknown protein sequence data is being generated and accumulated, leading to an increasing demand for protein classification and annotation. Existing machine learning methods still have limitations in protein sequence classification, such as low accuracy and precision of classification models, rendering them less valuable in practical applications. Additionally, these models often lack strong generalization capabilities and cannot be widely applied to various types of proteins. Therefore, accurately classifying and predicting proteins remains a challenging task. In this study, we propose a protein sequence classifier called Multi-Laplacian Regularized Random Vector Functional Link (MLapRVFL). By incorporating Multi-Laplacian and L2,1−norm regularization terms into the basic Random Vector Functional Link (RVFL) method, we effectively improve the model's generalization performance, enhance the robustness and accuracy of the classification model. The experimental results on two commonly used datasets demonstrate that MLapRVFL outperforms popular machine learning methods and achieves superior predictive performance compared to previous studies. In conclusion, the proposed MLapRVFL method makes significant contributions to protein sequence prediction.

Detection of COVID-19 Using Protein Sequence Data via Machine Learning Classification Approach

The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in late 2019 resulted in the COVID-19 pandemic, necessitating rapid and accurate detection of pathogens through protein sequence data. This study is aimed at developing an efficient classification model for coronavirus protein sequences using machine learning algorithms and feature selection techniques to aid in the early detection and prediction of novel viruses. We utilized a dataset comprising 2000 protein sequences, including 1000 SARS-CoV-2 sequences and 1000 non-SARS-CoV-2 sequences. Feature extraction provided 27 essential features representing the primary structural data, achieved through the Discere package. To optimize performance, we employed machine learning classification algorithms such as K-nearest neighbor (KNN), XGBoost, and Naïve Bayes, along with feature selection techniques like genetic algorithm (GA), LASSO, and support vector machine recursive feature elimination (SVM-RFE). The SVM-RFE+KNN model exhibited exceptional performance, achieving a classification accuracy of 99.30%, specificity of 99.52%, and sensitivity of 99.55%. These results demonstrate the model’s efficacy in accurately classifying coronavirus protein sequences. Our research successfully developed a robust classification model capable of early detection and prediction of protein sequences in SARS-CoV-2 and other coronaviruses. This advancement holds great promise in facilitating the development of targeted treatments and preventive strategies for combating future viral outbreaks.

Accurate Prediction of Protein Sequences for Proteogenomics Data Integration.

This book chapter discusses proteogenomics data integration and provides an overview into the different omics layer involved in defining the proteome of a living organism. Various aspects of genome variability affecting either the sequence or abundance level of proteins are discussed in this book chapter, such as the effect of single-nucleotide variants or larger genomic structural variants on the proteome. Next, various sequencing technologies are introduced and discussed from a proteogenomics data integration perspective such as those providing short- and long-read sequencing and listing their respective advantages and shortcomings for accurate protein variant prediction using genomic/transcriptomics sequencing data. Finally, the various bioinformatics tools used to process and analyze DNA/RNA sequencing data are discussed with the ultimate goal of obtaining accurately predicted sample-specific protein sequences that can be used as a drop-in replacement in existing approaches for peptide and protein identification using popular database search engines such as MSFragger, SearchGUI/PeptideShaker.

A Comparative Study of Protein Sequences Classification-Based Machine Learning Methods for COVID-19 Virus against HIV-1

ABSTRACT The effective spread of COVID-19 cases in several countries produces more protein sequences that are released in genomic public sources. It provides some awareness and indications for virus classification of COVID-19 and HIV-1 that are essential for drug discovery of COVID-19. This paper reveals the importance of machine learning algorithms to handle the recognition of two different viruses. Therefore, 18,476 protein sequences for both COVID-19 and HIV-1 and 9238 for each virus are applied to the proposed model based on feature extraction, data labeling, and six classifiers. Amino acid classification according to their dipoles and volumes is employed as a feature extraction tool based on the creation of eight features from twenty amino acids by using the conjoint triad (CT) method. The data labeling is employed as a coding tool by binary numbers refereeing zero for COVID-19 and one for HIV-1. The random forest (RF) model achieved the highest classification accuracy of 99.89% for eight features and 97.80% for two features. The experimental results significantly confirmed that eight features required more computational time than two features, but the accuracy rate was nearly similar in the two cases. This classification strategy of COVID-19 and HIV-1 will prompt the prediction of protein sequences of the new virus.

Improving Prediction of Helminth Aminoacyl‐tRNA Synthetases through Manual Analysis of Genomic Data

Helminth is a generic term for the parasitic worms that negatively impact health outcomes for a roughly a quarter of the global population. Although more than 150 helminth genomes have now been sequenced, protein sequence predictions available in compiled genomes and public databases are often incorrect. These errors hinder identification of novel drug targets in helminth genomes. We have focused on identifying and correcting predictions for aminoacyl-tRNA synthetase genes. These enzymes are targets of effective antiparasitic therapies in in other organisms. Using published genomic information available in WormBase from a set of twelve human-infecting helminths, we initially corrected exon boundary predictions using RNAseq data. The proposed protein sequences were then validated using multiple-sequence alignments of enzymes from helminths and other organisms, allowing us to identify helminth-specific features. We confirmed a high error rate in predictions of the studied genes, with 60% requiring at least one correction and an average of 4.3 corrections made per gene. Our analysis also identified several unique features of helminth aminoacyl-tRNA synthetases, particularly among trematode (fluke) and platyhelminth (flatworm) proteins and in mitochondrially-targeted enzymes.

Challenges of automation and scale: Bioinformatics and the evaluation of proteins to support genetically modified product safety assessments

Bioinformatic analyses of protein sequences play an important role in the discovery and subsequent safety assessment of insect control proteins in Genetically Modified (GM) crops. Due to the rapid adoption of high-throughput sequencing methods over the last decade, the number of protein sequences in GenBank and other public databases has increased dramatically. Many of these protein sequences are the product of whole genome sequencing efforts, coupled with automated protein sequence prediction and annotation pipelines. Published genome sequencing studies provide a rich and expanding foundation of new source organisms and proteins for insect control or other desirable traits in GM products. However, data generated by automated pipelines can also confound regulatory safety assessments that employ bioinformatics. Largely this issue does not arise due to underlying sequence, but rather its annotation or associated metadata, and the downstream integration of that data into existing repositories. Observations made during bioinformatic safety assessments are described.

Linoleic acid and α-linolenic acid inhibit conjugative transfer of an IncX4 plasmid carrying mcr-1.

The aim of this study was to determine the effects of unsaturated fatty acids on clinical plasmids. Two unsaturated fatty acids, linoleic acid (LA) and α-linolenic acid (ALA) at final concentration 0, 0·03, 0·3 and 3mmoll-1 , respectively, were used to assess the effects on conjugative transfer of a mcr-1-harbouring plasmid pCSZ4 (IncX4) in conjugation experiment. The inhibitory mechanisms were analysed by molecular docking and the gene expression of virB11 was quantitated by qRT-PCR. Target plasmid diversity was carried out by TrwD/VirB11 homology protein sequence prediction analysis. Our results showed that LA and ALA inhibit plasmid pCSZ4 transfer by binding to the amino acid residues (Phe124 and Thr125) of VirB11 with dose-dependent effects. The expression levels of virB11 gene were also significantly inhibited by LA and ALA treatment. Protein homology analysis revealed a wide distribution of TrwD/VirB11-like genes among over 37 classes of plasmids originated from both Gram-negative and Gram-positive bacteria. This study demonstrates representing a diversity of plasmids that may be potentially inhibited by unsaturated fatty acids. Our work reported here provides additional support for application of curbing the spread of multiple plasmids by unsaturated fatty acids.

Insights from the revised complete genome sequences of Acinetobacter baumannii strains AB307-0294 and ACICU belonging to global clones 1 and 2.

The Acinetobacter baumannii global clone 1 isolate AB307-0294, recovered in the USA in 1994, and the global clone 2 (GC2) isolate ACICU, isolated in 2005 in Italy, were among the first A. baumannii isolates to be completely sequenced. AB307-0294 is susceptible to most antibiotics and has been used in many genetic studies, and ACICU belongs to a rare GC2 lineage. The complete genome sequences, originally determined using 454 pyrosequencing technology, which is known to generate sequencing errors, were re-determined using Illumina MiSeq and MinION (Oxford Nanopore Technologies) technologies and a hybrid assembly generated using Unicycler. Comparison of the resulting new high-quality genomes to the earlier 454-sequenced versions identified a large number of nucleotide differences affecting protein coding sequence (CDS) features, and allowed the sequences of the long and highly repetitive bap and blp1 genes to be properly resolved for the first time in ACICU. Comparisons of the annotations of the original and revised genomes revealed a large number of differences in the protein CDS features, underlining the impact of sequence errors on protein sequence predictions and core gene determination. On average, 400 predicted CDSs were longer or shorter in the revised genomes and about 200 CDS features were no longer present.

Expression and Functional Analysis of the BCL2-Associated Agonist of Cell Death (BAD) Gene in the Sheep Ovary During the Reproductive Cycle.

Most ewes in China are seasonally polyestrous with normal ovulatory cycles, which is controlled by photoperiod (length of the daily light phase). These ewes are estrous in the short-day season and anestrus in the long-day season and cannot mate during anestrus. Thus seasonal breeding limits both diversification and intensification of production. If sheep can estrus all round year, it can be mated twice per year, which can greatly improve the economic benefits. To change seasonal estrus at the genetic level and cultivating new sheep breeds, it is important to understand the molecular mechanisms of seasonal breeding trait in sheep. The BCL2-associated agonist of cell death (BAD) gene being a regulator of cellular apoptosis was identified by our previous RNA-Seq, which is associated with follicular development in mammalian ovaries. However, the mechanism how BAD can regulate estrus in sheep was poorly understood. In this study, we characterized ovine BAD, including full-length mRNA cloning and protein sequence prediction, as well as BAD expression profile in Small-tailed Han (STH) sheep. The highest expression levels of BAD were observed in sheep hypothalamus, lung, and pituitary, while the lowest expression was in liver. Functional analysis of BAD was performed in primary granulosa cells of sheep. The concentration of P4 was significantly increased after RNAi interference of BAD, while P4 level was shown to be opposite after BAD overexpression in vitro. It has been found that BAD can reduce progesterone levels by promoting ovarian GC apoptosis, which might be involved in regulating the estrus cycle in sheep.

Large-scale automated function prediction of protein sequences and an experimental case study validation on PTEN transcript variants.

Recent advances in computing power and machine learning empower functional annotation of protein sequences and their transcript variations. Here, we present an automated prediction system UniGOPred, for GO annotations and a database of GO term predictions for proteomes of several organisms in UniProt Knowledgebase (UniProtKB). UniGOPred provides function predictions for 514 molecular function (MF), 2909 biological process (BP), and 438 cellular component (CC) GO terms for each protein sequence. UniGOPred covers nearly the whole functionality spectrum in Gene Ontology system and it can predict both generic and specific GO terms. UniGOPred was run on CAFA2 challenge target protein sequences and it is categorized within the top 10 best performing methods for the molecular function category. In addition, the performance of UniGOPred is higher compared to the baseline BLAST classifier in all categories of GO. UniGOPred predictions are compared with UniProtKB/TrEMBL database annotations as well. Furthermore, the proposed tool's ability to predict negatively associated GO terms that defines the functions that a protein does not possess, is discussed. UniGOPred annotations were also validated by case studies on PTEN protein variants experimentally and on CHD8 protein variants with literature. UniGOPred protein functional annotation system is available as an open access tool at http://cansyl.metu.edu.tr/UniGOPred.html.

Next-Step Conditioned Deep Convolutional Neural Networks Improve Protein Secondary Structure Prediction

Recently developed deep learning techniques have significantly improved the accuracy of various speech and image recognition systems. In this paper we show how to adapt some of these techniques to create a novel chained convolutional architecture with next-step conditioning for improving performance on protein sequence prediction problems. We explore its value by demonstrating its ability to improve performance on eight-class secondary structure prediction. We first establish a state-of-the-art baseline by adapting recent advances in convolutional neural networks which were developed for vision tasks. This model achieves 70.0% per amino acid accuracy on the CB513 benchmark dataset without use of standard performance-boosting techniques such as ensembling or multitask learning. We then improve upon this state-of-the-art result using a novel chained prediction approach which frames the secondary structure prediction as a next-step prediction problem. This sequential model achieves 70.3% Q8 accuracy on CB513 with a single model; an ensemble of these models produces 71.4% Q8 accuracy on the same test set, improving upon the previous overall state of the art for the eight-class secondary structure problem. Our models are implemented using TensorFlow, an open-source machine learning software library available at this http URL; we aim to release the code for these experiments as part of the TensorFlow repository.

OSPREY Predicts Resistance Mutations Using Positive and Negative Computational Protein Design.

Drug resistance in protein targets is an increasingly common phenomenon that reduces the efficacy of both existing and new antibiotics. However, knowledge of future resistance mutations during pre-clinical phases of drug development would enable the design of novel antibiotics that are robust against not only known resistant mutants, but also against those that have not yet been clinically observed. Computational structure-based protein design (CSPD) is a transformative field that enables the prediction of protein sequences with desired biochemical properties such as binding affinity and specificity to a target. The use of CSPD to predict previously unseen resistance mutations represents one of the frontiers of computational protein design. In a recent study (Reeve et al. Proc Natl Acad Sci U S A 112(3):749-754, 2015), we used our OSPREY (Open Source Protein REdesign for You) suite of CSPD algorithms to prospectively predict resistance mutations that arise in the active site of the dihydrofolate reductase enzyme from methicillin-resistant Staphylococcus aureus (SaDHFR) in response to selective pressure from an experimental competitive inhibitor. We demonstrated that our top predicted candidates are indeed viable resistant mutants. Since that study, we have significantly enhanced the capabilities of OSPREY with not only improved modeling of backbone flexibility, but also efficient multi-state design, fast sparse approximations, partitioned continuous rotamers for more accurate energy bounds, and a computationally efficient representation of molecular-mechanics and quantum-mechanical energy functions. Here, using SaDHFR as an example, we present a protocol for resistance prediction using the latest version of OSPREY. Specifically, we show how to use a combination of positive and negative design to predict active site escape mutations that maintain the enzyme's catalytic function but selectively ablate binding of an inhibitor.

ASAP: a machine learning framework for local protein properties.

Determining residue-level protein properties, such as sites of post-translational modifications (PTMs), is vital to understanding protein function. Experimental methods are costly and time-consuming, while traditional rule-based computational methods fail to annotate sites lacking substantial similarity. Machine Learning (ML) methods are becoming fundamental in annotating unknown proteins and their heterogeneous properties. We present ASAP (Amino-acid Sequence Annotation Prediction), a universal ML framework for predicting residue-level properties. ASAP extracts numerous features from raw sequences, and supports easy integration of external features such as secondary structure, solvent accessibility, intrinsically disorder or PSSM profiles. Features are then used to train ML classifiers. ASAP can create new classifiers within minutes for a variety of tasks, including PTM prediction (e.g. cleavage sites by convertase, phosphoserine modification). We present a detailed case study for ASAP: CleavePred, an ASAP-based model to predict protein precursor cleavage sites, with state-of-the-art results. Protein cleavage is a PTM shared by a wide variety of proteins sharing minimal sequence similarity. Current rule-based methods suffer from high false positive rates, making them suboptimal. The high performance of CleavePred makes it suitable for analyzing new proteomes at a genomic scale. The tool is attractive to protein design, mass spectrometry search engines and the discovery of new bioactive peptides from precursors. ASAP functions as a baseline approach for residue-level protein sequence prediction. CleavePred is freely accessible as a web-based application. Both ASAP and CleavePred are open-source with a flexible Python API.Database URL: ASAP’s and CleavePred source code, webtool and tutorials are available at: https://github.com/ddofer/asap; http://protonet.cs.huji.ac.il/cleavepred.

Genes encoding novel secreted and transmembrane proteins are temporally and spatially regulated during Drosophila melanogasterembryogenesis

BackgroundMorphogenetic events that shape the Drosophila melanogaster embryo are tightly controlled by a genetic program in which specific sets of genes are up-regulated. We used a suppressive subtractive hybridization procedure to identify a group of developmentally regulated genes during early stages of D. melanogaster embryogenesis. We studied the spatiotemporal activity of these genes in five different intervals covering 12 stages of embryogenesis.ResultsMicroarrays were constructed to confirm induction of expression and to determine the temporal profile of isolated subtracted cDNAs during embryo development. We identified a set of 118 genes whose expression levels increased significantly in at least one developmental interval compared with a reference interval. Of these genes, 53% had a phenotype and/or molecular function reported in the literature, whereas 47% were essentially uncharacterized. Clustering analysis revealed demarcated transcript groups with maximum gene activity at distinct developmental intervals. In situ hybridization assays were carried out on 23 uncharacterized genes, 15 of which proved to have spatiotemporally restricted expression patterns. Among these 15 uncharacterized genes, 13 were found to encode putative secreted and transmembrane proteins. For three of them we validated our protein sequence predictions by expressing their cDNAs in Drosophila S2R+ cells and analyzed the subcellular distribution of recombinant proteins. We then focused on the functional characterization of the gene CG6234. Inhibition of CG6234 by RNA interference resulted in morphological defects in embryos, suggesting the involvement of this gene in germ band retraction.ConclusionOur data have yielded a list of developmentally regulated D. melanogaster genes and their expression profiles during embryogenesis and provide new information on the spatiotemporal expression patterns of several uncharacterized genes. In particular, we recovered a substantial number of unknown genes encoding putative secreted and transmembrane proteins, suggesting new components of signaling pathways that might be incorporated within the existing regulatory networks controlling D. melanogaster embryogenesis. These genes are also good candidates for additional targeted functional analyses similar to those we conducted for CG6234.See related minireview by Vichas and Zallen:

Expression feature and bioinformatic analysis of molar root patterning gene 1

To examine the expression of molar root patterning gene 1 (Mrp1) and predict the Mrp1 structure by bioinformatics analysis. A pair of Mrp1-specific PCR primers were designed, and RT-PCR method was used to study the mRNA's expression pattern in rat molar root and other organs. Gene positioning and other protein sequence prediction were carried out by chromosome analysis and other bioinformatics analysis. Mrp1 was expressed not only in the molar but also in the developing pancreas, liver, lung and kidney tissues. Mrp1 was located in the 18q12.3 chromosome of the rats and the Mrp1 amino acids sequence had about 37% homology with a known protein Uroplakin IIIb (p35) which was an urothelial differentiation membrane molecular marker. A trans-membrane structure, 5 PKC phosphorylation sites and 4 CKII phosphorylation sites in Mrp1 were found. Mrp1 has a broad expression in different developing organs, and it may have a important function in the rat tooth root development.

Resolving the database sequence discrepancies for the Staphylococcus aureus bacteriophage ϕ 11 amidase

There are two conflicting primary nucleotide sequences of the Staphylococcus aureus bacteriophage phi 11 amidase gene in Genbank. Nucleotide sequence differences as well as alternative translational start site assignments result in three non-identical protein sequence predictions for this amidase. Therefore, it is prudent to verify the correct phi11 amidase protein sequence, especially since multiple versions of the amidase gene have been subcloned, deletion analysis performed, and their experimental use described. There is also a resurgence of interest in the expression and use of bacteriophage lytic proteins as bactericidal agents and the phi 11 amidase has a high antimicrobial potential. The correct amidase sequence is identified through a combination of DNA sequence analysis and matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry analysis of the recombinant purified phi11 amidase protein.

The universal algorithm of maturation for secretory and excretory protein precursors

During maturation, most proteins undergo different posttranslational modifications. In most simple cases, signal peptidases remove the signal or leader peptide from the precursors of the secretory proteins during their translocation across the ER membrane. For biologically active proteins, such as enzymes, regulatory and defense proteins, toxins, etc., additional maturation-regulating mechanisms were shown to proceed with limited proteolysis of inactive precursors by specific enzymes. A number of specific enzymes from different cell types selectively cleave proproteins at specific processing sites. In this work, we analyzed the sequences of protein precursors synthesized in the excretory glands of different animals and identified new, non-traditional processing sites. They differ from the motifs previously identified in secreted proteins’ precursors and enabled us to reconstruct the sequence of events leading to the conversion of protein precursors into the final products (mature proteins). We also found that in animals, the maturation mechanism of secretory and excretory proteins and the set of enzymes involved are species specific. The processing sites identified in protein precursors in this study are useful for a more detailed genome analysis and more accurate mature protein sequence prediction.

EST-PAC a web package for EST annotation and protein sequence prediction.

With the decreasing cost of DNA sequencing technology and the vast diversity of biological resources, researchers increasingly face the basic challenge of annotating a larger number of expressed sequences tags (EST) from a variety of species. This typically consists of a series of repetitive tasks, which should be automated and easy to use. The results of these annotation tasks need to be stored and organized in a consistent way. All these operations should be self-installing, platform independent, easy to customize and amenable to using distributed bioinformatics resources available on the Internet.In order to address these issues, we present EST-PAC a web oriented multi-platform software package for expressed sequences tag (EST) annotation. EST-PAC provides a solution for the administration of EST and protein sequence annotations accessible through a web interface. Three aspects of EST annotation are automated: 1) searching local or remote biological databases for sequence similarities using Blast services, 2) predicting protein coding sequence from EST data and, 3) annotating predicted protein sequences with functional domain predictions. In practice, EST-PAC integrates the BLASTALL suite, EST-Scan2 and HMMER in a relational database system accessible through a simple web interface. EST-PAC also takes advantage of the relational database to allow consistent storage, powerful queries of results and, management of the annotation process. The system allows users to customize annotation strategies and provides an open-source data-management environment for research and education in bioinformatics.