
==== Front
Mol Biol Res Commun
Mol Biol Res Commun
MBRC
Molecular Biology Research Communications
2322-181X
2345-2005
Shiraz University Shiraz, Iran

10.22099/mbrc.2024.50123.1981
Original Article
In-silico comparison of fungal and bacterial asparaginase enzymes
Tafvizi Negar
Behbahani Mandana
Mohabatkar Hassan *
Department of Biotechnology, Faculty of Biological Science and Technology, University of Isfahan, Isfahan, Iran
* Corresponding Author: Department of Biotechnology, Faculty of Biological Science and Technology, University of Isfahan, Isfahan, Iran. Tel: +98 313 793 4391; Fax: +98 313 793 2342; Email: h.mohabatkar@ast.ui.ac.ir
2024
13 4 183191
https://creativecommons.org/licenses/by/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/)
L-asparaginase is a commercial enzyme with a wide variety of applications. Asparaginase is known as an anti-cancer agent that is effective for the treatment of certain lymphomas and leukemias by growth inhibition of human cancer cells. Additionally, asparaginase is used in the food industry in a pretreatment process to decrease the accumulation of carcinogenic acrylamide. In this paper, different aspects of bacterial and fungal asparaginases such as mass, hydrophobicity and hydrophilicity of pseudo amino acid composition (PseAAC), physicochem-ical properties, and structural motifs were studied, and ROC curve statistical analysis was used for the comparison. The results showed that none of the physicochemical properties of fungal and bacterial asparaginase could not be differed, except molecular weight and sequence length. MEME Suite analysis demonstrated that there was a motif that was specific for bacterial asparaginases. However, analysis based on the concept of PseACC indicated a differentiation line between fungal and bacterial asparaginases. In conclusion, although there was not any specific demonstration to separate the bacterial and fungal asparaginases in the case of physicochemical properties, PseAAC analysis can be an appropriate and usable method to differentiate between them.

Key Words

L-asparaginase
Pseudo amino acid composition
Physicochemical properties
==== Body
pmcINTRODUCTION

L-asparaginase (L-Asparagine amidohydrolase, EC 3.5.1.1) catalyzes the hydrolysis of asparagine amino acid to aspartic acid and ammonia [1]. It can also break down glutamine into ammonia and glutamic acid. Asparaginase has been found in tetrameric, dimeric, hexameric, and monomeric forms when extracted from different sources [2]. In addition, it was identified that there were two isozymes of asparaginase called type 1 and type 2 [3]. However, type 2 has a higher affinity for asparagine than type 1 [4]. Studies have reported the presence of asparaginase in various sources including animals, plants, and microorganisms (bacteria, fungi, algae, and yeasts). In recent years, many papers have focused on this enzyme’s ability because of its biotechnological applications and simplicity of large-scale production [5].

The first reason for asparaginas importance is the antineoplastic property that scientists and researchers broadly have observed it strong. Neoplastic cells cannot produce asparagine for their metabolic needs due to the low expression or absence of the asparagine synthetase gene. Therefore, they provide their required asparagine from the surrounding environment. In this regard, asparaginase is an induction chemotherapy standard treatment option against tumor cell growth for acute lymphocytic leukemia (ALL) [2, 5]. The main microbial asparaginase resources for medical approaches are Escherichia sp. and Erwinia sp [1].

L-asparaginase also has a potential role in the food industry as a food processor [6]. Acrylamide is a suspected carcinogen, made in cooked food, and asparaginase is used to reduce acrylamide formation [7]. Asparaginases from diverse sources such as Aspergillus niger and A. oryzae have been assayed in acrylamide reduction in different foods [8].

Regarding asparaginase applications in therapy and biotechnology, there is a problem that should be considered. Prokaryotic resources of asparaginase production can have side effects in long-term therapy including hypersensitivity and immune reactions [9]. To solve this problem, finding new eukaryotic sources of asparaginase production can be a considerable solution.

According to the solution mentioned above, it is important to know which asparaginase source is superior in fungi and bacteria based on physicochemical features. Nowadays, bioinformatics is developed to help researchers with the classification and prediction of various aspects of enzymes to save time and cost.

Several studies about amino acid sequence and protein structure analysis of enzymes have been reported in the last decade. The amino acid composition of L-asparaginases from fungi, bacteria, and plants was widely studied and compared to differentiate each asparaginase sequence [10-11]. In-silico characterization of fungal asparaginase sequences as well as biochemical characters of the purified enzymes was investigated [12]. Physicochemical properties of endophytic bacterial asparaginase enzymes were characterized by some bioinformatics tools [13]. In the present study, in-silico characterization of fungal asparaginases were compared with bacterial asparaginases for the first time.

MATERIALS AND METHODS

Datasets of L-asparaginase: Protein sequences of bacterial and fungal asparaginase were selected from NCBI (https://www.ncbi.nlm.nih.gov). 786 and 8342 asparaginase protein sequences from fungi and bacteria, were chosen as our datasets, respectively. To cluster repeated and similar sequences and choose a representative sequence, the CD-HIT program (http://cd-hit.org) and Decrease Redundancy sever from Expasy (https://web.expasy.org) with 95% sequence identity cutoff were employed. The final results included 281 and 229 bacterial and fungal sequences, respectively.

Physicochemical parameters analysis: Physicochemical properties of primary protein sequences of fungal and bacterial asparaginases were predicted by ProtParam tool (https://web.expasy.org). ProtParam is available on Expasy website. The properties consisted of sequence length, molecular weight, theoretical PI, aliphatic index, instability index, positively charged, negatively charged, and grand average of hydropathicity (GRAVY).

Pseudo amino acid compositions (PseACC): Identifying various features of uncharacterized proteins is one of the most important tasks facing us today in bioinformatics because its obtained information has a significant effect on the improvement of system’s biology and proteomics [14]. That’s why the concept of PseACC was introduced in 2001 [15]. PseACC is based on both amino acid composition and sequence order effect. This method is introduced to extract the features of amino acid composition and physical and chemical characteristics of amino acids [16]. In another word, PseACC is described by a set of 20+λ distinct factors. The 20 part shows the AAC components and the λ part illustrates sequence order correlation [17]. PseACC plays an important role in converting a character sequence of a protein to a numerical sequence. The feature descriptor extraction and model construction were implemented in a machine-learning platform that served as protein sequence analysis and prediction [18]. Different physicochemical parameters like hydrophobicity, hydrophilicity, and side chain mass are the factors to make such a conversion [19].

In this study, BioSeq-Analysis 2.0 web server (http://bliulab.net/BioSeq-Analysis2.0/) was utilized to achieve the numerical sequence. Four machine learning algorithms with diverse component modes were analyzed. The first machine learning algorithm was support vector machine; a powerful classification method based on the idea of the generalized linear classifier [20]. OET-KNN algorithm was the second one, worked based on the Dempster-Shafer theory. Each neighbor in a pattern was classified as evidence supporting certain hypotheses [21]. Another machine was Random Forest which represented randomly generated trees, predicted different classes as a training dataset, and then the class with a greater predicted number among the trees was implemented as the test data [22]. The last platform was Covariance Discriminant machine-learning. This analysis was the formulation of classifying rules based on multiple training datasets which are classified by those determined rules [23]. The results of the assessment were evaluated via five parameters: ACC (Overall accuracy), MCC (Mathews Correlation Coefficient), AUC (Area under the curve), Sn (Sensitivity) and Sp (Specificity). ACC, MCC, Sn and Sp were calculated referring to Eqs (1-4) [17].

Acc = (TP+TN)/(TP+TN+FP+FN)

MCC = ((TP*TN) – (FP*FN))/√((TP+FP) (TP+FN) (TN+FP) (TN+FN))

Sn = TP/(TP+FN)

Sp = TN/(TN+FP)

TP, TN, FP and FN are, respectively short form of True Positive, True Negative, False Positive and False Negative.

Secondary structure analysis: It is proved that structural information provides insight into protein function [24]. In this regard, we can employ automatic database search methods to design drugs and understand more details of protein-protein interaction networks [25]. For predicting the secondary structure of the protein, the Garnier-Osguthorpe-Robson IV (GOR IV) online tool (https://npsa-prabi.ibcp.fr/) was used [26] ,and Alpha helix, extended strand, and random coiled were predicted.

Motif discovery: The conserved motifs in bacterial and fungal asparaginase were predicted by the MEME Suite web server (http://meme.nbcr.net). Analysis of motif sequences represents features like DNA binding sites and protein interaction domains [27]. For predicting domains and the family relations of characterized motifs, InterPro (https://www.ebi.ac.uk) was employed [28]. Here, conserved motifs in both bacterial and fungal asparaginases were compared.

Statistical analysis: Statistical tests were introduced to compare two or more different diagnostic systems [29]. To compare amino acid composition and secondary structure physicochemical properties of asparaginases between bacteria and fungi sequences, Receiver Operator Characteristic (ROC) curve analysis (http://melolab.org/star/roc_analysis.php) was used. Accuracy (ACC) is the parameter that describes a binary classification. The acceptable classification performance is described when ACC is more than 80%. Bio-seq Analysis 2.0 server also uses the ROC analysis system for comparison.

RESULTS

To evaluate the results of ProtParam server, ROC curve was used (Table 1). Two parameters, sequence length and molecular weight, showed the differences between bacterial and fungal enzymes receiving more than 80% accuracy. However, other physiochemical properties did not show significant differences. In addition, the frequency of all 20 amino acids in bacterial and fungal asparaginases was determined (Table 2). ROC curve indicated that bacterial asparaginases could be distinguishable from fungal asparaginases in aspect of amino acid frequency such as glutamic acid, glycine, cysteine, and serine.

Table 1 ROC curve analysis of ProtParam results

Physico-chemical parameters	ACC	
Theoretical PI	0.6503	
Instability index	0.6523	
Sequence length	0.9198	
Molecular weight	0.8959	
Aliphatic index	0.6405	
GRAVY	0.5658	
(The acceptable results were shown in bold)

Table 2 ROC analysis of amino acids frequency between bacterial and fungal asparaginases

Amino acids	ACC	
Alanine	0.7390	
Arginine	0.7217	
Asparagine	0.5797	
Aspartic acid	0.6027	
Cysteine	0.8503	
Glutamine	0.5739	
Glutamic acid	0.8292	
Glycine	0.8484	
Histidine	0.6679	
Iso leucine	0.7044	
leucine	0.7447	
Lysine	0.6833	
Methionine	0.5854	
Phenyl alanine	0.7639	
Proline	0.6718	
Serine	0.8292	
Threonine	0.6180	
Tryptophan	0.4395	
Tyrosine	0.6641	
Valine	0.6814	
(The acceptable results were shown in bold)

For analyzing the results of PseACC, four types of machine learning algorithms while considering three physicochemical parameters (mass, hydrophobicity, hydrophilicity), were used (Table 3). The results showed that bacterial and fungal Asparaginase sequences were significantly different. The highest performance was achieved by OET-KNN machine-algorithm (ACC 99.06%).

The results of secondary structure prediction were evaluated using ROC curve analysis (Table 4). In the case of the secondary structure prediction, the results exhibited that bacterial Asparaginases were not significantly different from fungal asparaginases. ACC values of all three parameters were less than 80%.

MEME suite server highlighted three bacterial (Motif A, B, and C) and three fungal asparaginase motifs (Motif A, B, and C). InterPro server expressed family relations and length of bacterial and fungal asparaginase motifs, which are shown in Figures 1 and 2. On average, fungal motifs were shorter than bacterial ones. Motifs A and B of bacterial asparaginases belonged to the same family type as Motifs A and B of fungal sequences. Motif C of bacterial sequences was a member of asparaginase/glutaminase-like. While fungal Motif C was characterized as a member of the asparaginase II family.

Table 3 The results of four different machine-learning algorithm

Random Forest machine-learning algorithm	
ACC	0.9891	
MCC	0.9777	
AUC	0.9997±0.0004	
Sn	0.9858	
Sp	0.9935	
Support vector machine-learning algorithm	
ACC	0.9784	
MCC	0.9551	
AUC	0.9957±0.0033	
Sn	0.9825	
Sp	0.9724	
Covariance Discriminant machine-learning algorithm	
ACC	0.9555	
MCC	0.9137	
AUC	0.2904±0.054	
Sn	0.9235	
Sp	1.0	
OET-KNN machine-learning algorithm	
ACC	0.9906	
MCC	0.9808	
AUC	0.9978±0.0028	
Sn	0.991	
Sp	0.9901	
(The acceptable results were shown in bold)

Table 4 ROC analysis of GOR IV server results between bacterial and fungal asparaginase sequences

parameters	ACC	
Alpha helix	0.7894	
Extended strand	0.6978	
Random coiled	0.6777	

Figure 1 The most probable motifs in bacterial sequences. Motif A. 50 amino acids; Family type: Asparaginase/glutaminase-like; Domain type: Asparaginase, N-terminal; Homologous Superfamily: Asparaginase/ glutaminase-like, Asparaginase, N-terminal domain superfamily; Active site type: Asparaginase/glutaminase, active site 2. Motif B. 50 amino acids; Family type: Asparaginase/glutaminase-like; Domain type: Asparaginase, C-terminal; Homologous Super Family: Asparaginase/glutaminase-like, Asparaginase, C-terminal. Motif C. 41 amino acids, Family type: Asparaginase/glutaminase-like; Homologous Superfamily: Asparaginase, N-terminal domain superfamily.

Four physicochemical properties (theoretical PI, instability index, aliphatic index, and Gravy) of all three types of enzyme motifs in bacteria and fungi were selected and analyzed by ROC analysis. The results shown in Table 5, indicated that bacterial asparaginase Motif C was different from fungal asparaginase Motif C in the case of aliphatic index, GRAVY, theoretical PI, and instability index. ROC analysis and InterPro results both showed the difference between bacterial and fungal asparaginases in Motif C. Therefore, it was concluded that the results of ROC analysis were correlated with InterPro results.

Figure 2 The most probable motifs in fungal sequences. Motif A. 23 amino acids, Family type: Asparaginase/ glutaminase-like; Domain type: Asparaginase, N-terminal; Homologous superfamily: Asparaginase/glutaminase-like. Motif B. 36 amino acids, Family type: Asparaginase/glutaminase-like; Domain type: Asparaginase, N-terminal; Homologous superfamily: Asparaginase/glutaminase-like. Motif C. 50 amino acids, Family type: Asparaginase-II.

Table 5 ROC analysis of physicochemical properties of motifs A, B and C

Physico-chemical properties	Motif A (ACC)	Motif B (ACC)	Motif C (ACC)	
Aliphatic index	0.5135	0.6850	0.8421	
Instability index	0.3321	03223	0.8421	
Theoretical PI	0.6679	0.6286	0.8483	
GRAVY	0.6750	0.6777	0.8483	

Discussion

The purpose of this study was to compare bacterial and fungal asparaginase’s different cases. Bacterial asparaginases were compared to fungal asparaginases by computational methods for the first time. In this regard, powerful bioinformatics techniques such as physicochemical properties, secondary structure, PseAAC, and motif analysis tools were used.

Our results exhibited a separation line between bacterial and fungal asparaginase sequences in some physicochemical parameters. The Molecular weight and sequence length of asparaginase between fungi and bacteria showed a critical difference. Since fungi being eukaryotes, it is expected that fungal asparaginase sequences are heavier, and longer than bacterial sequences, and our efforts showed the same outcome. Amino acid composition analysis of bacterial and fungal asparaginases expressed a significant difference in the frequency of a few amino acids such as glutamic acid, glycine, serine, and cysteine. The Aliphatic index is the parameter that demonstrates the volume of aliphatic side chains (alanine, valine, isoleucine, and leucine) that are present in a protein. The aliphatic index of fungal asparaginase motifs was greater than bacterial asparaginase motifs. Because the aliphatic index could be an effective factor for the increase of thermostability of the protein, fungal asparaginases may be more practical than bacterial asparaginases in industrial applications. Regarding the instability of the protein, if the index is smaller than 40, it means that the protein is stable which is the state that the optimal activity could be seen [30]. Our results showed that there was no difference in the case of stability between fungal asparaginases and bacterial asparaginases. GRAVY is considered as hydropathy value of amino acid composition. When this index is negative, the peptide is hydrophobic [31]. In bacteria, the number of hydrophobic asparaginases was more than in hydrophilic ones. But, in fungi, it was inverted. However, ROC analysis received an accuracy of 56.58%. According to the motif’s investigation outputs, the one and only difference between bacterial and fungal sequences was seen in motif C with a low accuracy (84.21%).

Previous studies have proved that Chou’s PseAAC was a useful technique to classify proteins. In 2017, the dissimilarity between Reverse transcriptase enzyme of HIV-1 and HIV-2 was explained using statistical analysis of PseAAC [32]. In one paper, the result of Lipase enzyme classification between two groups of bacteria and fungi presented over 80% accuracy based on the concept of PseAAC [33]. Sudheer Gupta and his colleagues compared the toxicity of peptides and proteins with various prediction models in the case of PseACC [34]. In 2019, Random Forest machine learning was examined as a model for predicting the difference between toxin and non-toxin peptides of animal [35]. In a published paper in 2020, the PseAAC of Alkaline phosphatase in one class and Acid phosphatase in another class was analyzed [36]. Our results confirmed the efficiency of PseACC technique to achieve more than 95% accuracy regarding three physicochemical features (mass, hydrophobicity, hydrophilicity) for all four machine algorithms. While, in other analyzed severs, there was not such accuracy that was revealed from BioSeq-Analysis server. This represents that using Chou’s PseAAC helps us to predict whether a given asparaginase sequence is related to bacteria or fungi.

These days, our concern is to find asparaginase from eukaryotic sources with lower toxicity, adverse effects, higher activity levels, and large-scale production to replace bacterial sources usage for therapeutic and industrial purposes. Our bioinformatical analysis showed differences in some ways particularly the BioSeq-Analysis outcomes that surprised us in the way of its accurate precision. In the future, our efforts can go in a way that analyzes more aspects of the asparaginase by computational techniques with higher accuracy.

Acknowledgment:

The authors of this paper acknowledge the University of Isfahan for the financial support of this research.

Conflict of Interest:

There is no conflict of interest associated with this paper.

Authors’ Contribution:

MB and HM planed and designed the presented idea. NT performed the computation and verified the analytical method then wrote the manuscript. All authors contributed to the final version of the manuscript.
==== Refs
References

1 Darvishi F Faraji N Shamsi F Production and structural modeling of a novel asparaginase in Yarrowia lipolytica Int J Biol Macromol 2019 125 955 961 30576739
2 Batool T Makky EA Jalal M Yusoff MM A comprehensive review on L-asparaginase and its applications Appl Biochem Biotechnol 2016 178 900 923 26547852
3 Ohnuma T Bergel F Bray RC Enzymes in cancer: asparaginase from chicken liver Biochem J 1967 103 238 245 6033763
4 Campell HA Mashburn LT Boyse EA Old LJ Two L-asparaginases from Escherichia coli B Their separation, purification, and antitumor activity Biochemistry 1967 6 721 730 5337885
5 Souza PM de Freitas MM Cardoso SL Pessoa A Guerra ENS Magalhaes PO Optimization and purification of L-asparaginase from fungi: A systematic review Crit Rev Oncol Hematol 2017 120 194 202 29198332
6 Dhanam JG Kannan S L-asparaginase-Types, perspectives and applications Advanced BioTech 2013 13 01 05
7 Cachumba JJM Fernandes Antunes FA Dias Peres GF Brumano LP Dos Santos JC Da Silva SS Current applications and different approaches for microbial L-asparaginase production Braz J Microbiol 2016 47 77 85 27866936
8 Shi R Liu Y Mu Q Jiang Z Yang S Biochemical characterization of a novel L-asparaginase from Paenibacillus barengoltzii being suitable for acrylamide reduction in potato chips and mooncakes Int J Biol Macromol 2017 96 93 99 27919811
9 Duval M Suciu S Ferster A Rialland X Nelken B Lutz P Benoit Y Robert A Manel AM Vilmer E Otten J Philippe N Comparison of Escherichia coli–asparaginase with Erwinia-asparaginase in the treatment of childhood lymphoid malignancies: results of a randomized European Organisation for Research and Treatment of Cancer-Children's Leukemia Group phase 3 trial Blood 2002 99 2734 2739 11929760
10 Mousavi SE Mohabatkar H Behbahani M Comparative in silico analysis of fungal and bacterial alkaline serine proteases: Insights into structure, function, and evolution Iran J Sci 2024 48 1 8
11 Dwivedi VD Mishra SK In silico analysis of L-asparaginase from different source organisms Interdiscip Sci 2014 6 93 99 25172447
12 Elsaba YM Salama WH Soliman ERS In silico and biochemical analysis on a newly isolated Trichoderma asperellum l-asparaginase Biocatal Agricul Biotech 2022 40 102309
13 Zadeh Hosseingholi E Neghabi N Molavi G Gheibi Hayat SM Shahriarpour H In silico identification and characterization of antineoplastic asparaginase enzyme from endophytic bacteria IUBMB Life  2020 72 991 1000 31981306
14 Chou KC Shen HB Recent advances in developing web-servers for predicting protein attributes Nat Sci 2009 1 63 92
15 Chou KC Prediction of protein cellular attributes using pseudo‐amino acid composition Proteins 2001 43 246 255 11288174
16 Su W Qian X Yang K Ding H Huang C Zhang Z Recognition of outer membrane proteins using multiple feature fusion Front Genet 2023 14 1211020 37351347
17 Samman N Mohabatkar H Rabiei P Using several pseudo amino acid composition types and different machine learning algorithms to classify and predict archaeal phospholipases Mol Biol Res Commun 2023 12 117 126 37525666
18 Huang A Lu F Liu F Discrimination of psychrophilic enzymes using machine learning algorithms with amino acid composition descriptor Front Microbiol 2023 14 1130594 36860491
19 Shen HB Chou KC PseAAC: a flexible web server for generating various kinds of protein pseudo amino acid composition Anal Biochem 2008 373 386 388 17976365
20 Manavalan B Basith S Shin TH Wei L Lee G mAHTPred: a sequence-based meta-predictor for improving the prediction of anti-hypertensive peptides using effective feature representation Bioinformatics 2019 35 2757 2765 30590410
21 Sen TZ Jernigan RL Garnier J Kloczkowski A GOR V server for protein secondary structure prediction Bioinformatics 2005 21 2787 2788 15797907
22 Alballa M Butler G Integrative approach for detecting membrane proteins BMC Bioinformatics 2020 21 575 33349234
23 Boulesteix AL Janitza S Kruppa J König IR Overview of random forest methodology and practical guidance with emphasis on computational biology and bioinformatics Wiley Interdiscip Rev Data Min Knowl Discov 2012 2 493 507
24 Sajana  OK Sajesh  TA Robust quadratic discriminant analysis using Sn covariance Commun Stat-Simul C 2023 52 735 744
25 Rost B Protein secondary structure prediction continues to rise J Struct Biol  2001 134 204 218 11551180
26 Garnier J Gibrat JF Robson B GOR method for predicting protein secondary structure from amino acid sequence Methods Enzymol 1996 266 540 553 8743705
27 Bailey TL Boden M Buske FA Frith M Grant CE Clementi L Ren J Li WW Noble WS MEME SUITE: tools for motif discovery and searching Nucleic Acids Res 2009 37 suppl_2 W202 W208 19458158
28 Mulder N Apweiler R Interpro and interproscan, in Comparative genomics Springer 2007 396 59 70
29 Hajian-Tilaki K Receiver operating characteristic (ROC) curve analysis for medical diagnostic test evaluation Caspian J Intern Med 2013 4 627 635 24009950
30 Ikai A Thermostability and aliphatic index of globular proteins J Biochem 1980 88 1895 1898 7462208
31 Lebreton A Moreau V Lapalud P Cayzac C André S Nguyen C Schved JF Lavigne G Granier C Discontinuous epitopes on the C2 domain of coagulation Factor VIII mapped by computer‐designed synthetic peptides Br J Haematol 2011 155 487 497 21933172
32 Behbahani M Mohabatkar H Nosrati M Discrimination of HIV-1 and HIV-2 reverse transcriptase proteins using Chou’s PseAAC Iran J Sci Technol, Trans A: Science 2017 42 1805 1811
33 Behbahani M Mohabatkar H Nosrati M Analysis and comparison of lignin peroxidases between fungi and bacteria using three different modes of Chou’s general pseudo amino acid composition J Theor Biol 2016 411 1 5 27615149
34 Gupta S Kapoor P Chaudhary K Gautam A Kumar R Open Source Drug Discovery Consortium Raghava GPS In silico approach for predicting toxicity of peptides and proteins PlOS One 2013 8 e73957 24058508
35 Pan Y Wang S Zhang Q Lu Q Su D Zuo Y Yang L Analysis and prediction of animal toxins by various Chou's pseudo components and reduced amino acid compositions J Theor Biol 2019 462 221 229 30452961
36 Amoozadeh M Behbahani M Mohabatkar H Keyhanfar M Analysis and comparison of alkaline and acid phosphatases of Gram-negative bacteria by bioinformatic and colorimetric methods J Biotech 2020 308 56 62
