Bulanık K-En Yakın Komşuluk Algoritmasında Lempel-Ziv Mesafe Ölçütünün Etkisi: Büyüme Faktörlerinin Sınıflandırılması Örneği

Year 2024, Volume: 5 Issue: 2, 148 - 162

Berk Tolga Çifci , Ramazan Kabadayı , Çağın Kandemir Çavaş

https://doi.org/10.53525/jster.1573661

Abstract

Hücresel olaylar, proteinlerin eylemleri sonucunda gerçekleşir. Amino asitlerin farklı dizilimleri farklı protein yapılarının oluşmasına neden olur. Yapılarına göre hücresel olaylardaki aktiviteleri de değişiklik gösterir. Bu nedenle protein dizilerinin yapısal veya işlevsel olarak sınıflandırılması hücresel olaylardaki rolleri hakkında bilgi edinmek için oldukça değerlidir. Büyüme faktörleri; hücreler üzerinde çoğalma, farklılaşma, onarım ve bakım gibi birçok süreçte yer alan proteinlerdir. Büyüme faktörlerinin in vivo çalışmaları kısa yarı ömre, zayıf bir dayanıklılığa yol açar. Biyoenformatik alanı temelinde literatürde NGF ve BDNF’nin sınıflandırılmasıyla ilgili herhangi bir çalışma bulunmamaktadır Büyüme faktörlerinin biyoenformatik alanında incelenmesi düşük maliyetle, daha hızlı sonuçlara ulaşılmasını sağlayabilir. Nörotrofinler; sinir hücrelerinin büyümesi, çoğalması, farklılaşması ve fonksiyonları üzerinde etkili olan büyüme faktörü ailelerinden biridir. Çalışmalar, her ne kadar nörotrofin ailesinin üyeleri olan NGF ve BDNF’ye dair bilgiler sunsa da hücresel ve moleküler işlevlerinin hala iyi anlaşılmadığını da göstermektedir. Biyoenformatik alanında yaygın olarak kullanılan k-En Yakın Komşuluk (KNN) algoritmasının performansı önemli ölçüde kullanılan mesafeye bağlıdır. Bulanık KNN (FKNN) algoritması için de mesafe ölçümleri, bulanıklık derecesini hesaplamak için önemlidir. Çalışmamızda, ortak bir atadan gelen ve çok benzer yüksek dereceli protein yapısına sahip olan NGF ve BDNF’nin, ayrıca NT-3’ün bulanık sınıflandırılması yapılmaktadır. Ayrıca çalışmada, FKNN algoritmasında test verisi ile eğitim verileri arasındaki mesafeyi ölçmek için protein sekanslarının Lempel-Ziv karmaşıklık değerlerine dayalı mesafe ölçümünün kullanılması önerilmektedir. Uniprot veri tabanından alınan verilerle birlikte FKNN algoritmasında Lempel-Ziv uzaklığı kullanıldığında K komşu sayısının 12 olması karşılığında, sınıflandırma performansı %83 olarak elde edilmiştir. Öklid Uzaklığı kullanıldığında elde edilen en yüksek sınıflandırma performansı ise %75’tir. Maksimum doğruluk oranını elde ettiğimiz noktada Öklid uzaklığını kullandığımızda algoritmamızın çalışma süresi 0.0054 ms iken Lempel-Ziv uzaklığı kullandığımızda 0.0038 ms’dir. Literatürde NGF ve BDNF’nin sınıflandırılmasıyla ilgili herhangi bir çalışma bulunmaması sebebiyle, elde edilen bulgular, makine öğrenmesi tekniklerinin nörotrofinlerin sınıflandırılmasında ilk kez uygulanması açısından bir yenilik sunmaktadır.

Keywords

Bulanık k-en yakın komşu algoritması, büyüme faktörü, Lempel-Ziv karmaşıklığı, nörotrofinler, protein sekansı.

The Impact of Lempel-Ziv Distance Metric in Fuzzy K-Nearest Neighbor Algorithm: A Case Study on Classification of Growth Factors

Abstract

Cellular events occur as a result of the actions of proteins. Different sequences of amino acids cause different protein structures. Their activities in cellular events also vary according to their structures. Therefore, structural or functional classification of protein sequences is very valuable for obtaining information about their role in cellular events. Growth factors that are proteins involved in many processes such as proliferation, differentiation, repair and maintenance on cells. In vivo studies of growth factors lead to short half-life, poor stability. Examination of growth factors in the field of bioinformatics can provide faster results at low cost. Neurotrophins are the one of the growth factor families that affect the growth, proliferation, differentiation and functions of nerve cells. Although studies provide information about NGF and BDNF, members of the neurotrophin family, they also show that their cellular and molecular functions are still not well understood. The performance of the k-nearest neighbor (KNN) algorithm, which is widely used in the field of bioinformatics, is significantly dependent on the distance used. For the fuzzy KNN algorithm (FKNN), distance measurements are important for calculating the degree of turbidity. In our study, fuzzy classification of NGF and BDNF, which comes from a common ancestor and has very similar high-grade protein structure, as well as NT-3 is made. In addition, in the study, it is recommended to use distance measurement based on Lempel-Ziv complexity values of protein sequences to measure the distance between test data and training data in FKNN algorithm. When the Lempel-Ziv distance was used in the FKNN algorithm with data from the Uniprot database, the classification performance was obtained as 83%, given that the number of K neighbors was 12. The highest classification performance achieved when Euclidean Distance is used is 75%. At the point where we obtain the maximum accuracy rate, the running time of our algorithm is 0.0054 ms when we use the Euclidean distance, while it is 0.0038 ms when we use the Lempel-Ziv distance. Since there is no study on the classification of NGF and BDNF in the literature, the findings provide an innovation in terms of the first application of machine learning techniques in the classification of neurotrophins.

Keywords

: Fuzzy K-nearest Neighbor Algorithm, Growth Factor, Lempel-Ziv Complexity, Neurotrophin, Protein Sequence

References

• [1] “Protein structure”, nature.com, 2014. [Online]. Available: https://www.nature.com/scitable/topicpage/protein-structure-14122136/. [Accessed: 6 June 2022].
• [2] K. Ahern, I. Rahagopal, T. Tan, “2.3: Structure & fuction- proteins I”, bio.libretext.org, Mar. 7, 2022. [Online]. Available: https://bio.libretexts.org/Bookshelves/Biochemistry/Book%3A_Biochemistry_Free_For_All_(Ahern_Rajagopal_and_Tan)/02%3A_Structure_and_Function/203%3A_Structure__Function-_Proteins_I [Accessed: June. 6, 2022]
• [3] J. Maillo, J. Luengo, S. Garcia, F. Herrera, I. Triguero, “Exact fuzzy k-nearest neighbor classification for big datasets”, 2017 IEEE International Conference on Fuzzy Systems (FUZ-IEEE), July 09-12, 2017, Naples, Italy [Online]. Available: IEEE Xplore, https://ieeexplore.ieee.org/document/8015686/authors#authors, [Accessed: 12 June 2022]
• [4] James M. Keller, Michael R. Gray, James A. Givens, “A fuzzy k-nearest neighbor algorithm”, IEEE Transactions on Systems, Man, and Cybernetics, vol: SMC-15, issue:4, pp. 580-585, July-Aug 1985, Doi: 10.1109/TSMC.1985.6313426. [Accessed: 15 June 2022]
• [5] X. Zheng, C. Li, J. Wang, “An information-theoretic approach to the prediction of protein structural class”, Journal of Computational Chemistry, vol. 31, issue 6, pp. 1201-1206, September 2009, Doi: 10.1002/jcc.21406. [Accessed: 28 June 2022]
• [6] JY. Chang, JJ. Shyu, YX. Shi (2008). “Fuzzy k-nearest neighbor classifier to predict protein solvent accessibility” Ishikawa, M., Doya, K. Miyamoto, H., Yamakawa, T., Neural Information Processing. ICONIP 2007, vol 4985, pp. 837-845, Springer, Berlin, Heidelberg. [Online]. Doi: https://doi.org/10.1007/978-3-540-69162-4_87. [Accessed: 28 June 2022]
• [7] Y. Huang, Y. Li, “Prediction of protein subcellular locations using fuzzy k-NN method”, Bioinformatics, vol. 20, no. 1, pages. 21-8, 2004 Jan. [Online]. Doi: 10.1093/bioinformatics/btg366. [Accessed: 25 June 2022]
• [8] R. Bondugula, O. Duzlevski, D. XU, “Profiles and fuzzy k-nearest neighbor algorithm for protein secondary structure prediction”, Proceedings of the 3rd Asia-Pacific Bioinformatics Conference, pp. 85-94, January 2005, Singapore, [Online]. Doi: 10.1142/9781860947322_0009. [Accessed: 1 July 2022]
• [9] M. Kumar, SK. Rath, “Microarray data classification using fuzzy k-nearest neighbor”, International Conference on Contemporary Computing and Informatics (IC3I), Mysore, India, IEEE, pp. 1032-1038, November 2014, Doi: 10.1109/IC3I.2014.7019618 [Accessed: 1 July 2022]
• [10] D. Li, JS, Deogun, K. Wang, “Gene function classification using fuzzy k-nearest neighbor approach”, 2007 IEEE International Conference on Granular Computing (GRC 2007), Fremont, CA, USA, IEEE Xplore, pp. 644-644, November 2007, Doi: 10.1109/GrC.2007.99. [Accessed: 28 June 2022]
• [11] J. Sim, SY. Kim, J. Lee, “Prediction of protein solvent accessibility using fuzzy k-nearest neighbor method”, Bioinformatics, vol. 21, issue 12, pages 2844-2849, April 2002 [Online] Doi: http://doi.org/10.1093/bioinformatics/bti423. [Accessed: 25 June 2022]
• [12] HA. Abu Alfeilat, ABA. Hassanat, O. Lasassmed, AS. Tarawneh, MB. Alhasanat, HS. Eyal Salman, VBS. Prasath, “Effects of distance measure choice on k-nearest neighbor classifier performance: a review”, Big Data, 7(4): 221-248, Dec 2019, Doi: 10.1089/big.2018.0175. Epub 2019 Aug 14. [Accessed: 8 July 2022]
• [13] K. Chomboon, P. Chujai, P. Teerarassamee, K. Kerdprasop, N. Kerdprasop, “An empirical study of distance metrics for k-nearest neighbor algorithm”, Proceedings of the 3rd International Conference on Industrial Application Engineering 2015, Japan, pp. 280-285, Doi: 10.12792/iciae2015.051. [Accessed: 5 July 2022]
• [14] P. Melin, E. Ramirez, G. Prado- Arechiga, “A new variant of fuzzy k-nearest neighbor using interval type-2 fuzzy logic”, 2018 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE), 2018, pp. 1-7, Doi: 10.1109/FUZZ-IEEE.2018.8491472. [Accessed: 5 July 2022]
• [15] PK. Jena, S. Chattopadhyay, “Comparative study of fuzzy k-nearest neighbor and fuzzy c-means algorithms”, International Journal of Computer Applications, vol. 57, no. 7, pp. 22-32, November 2012, Doi: 10.1007/978-3-642-30157-5_45. [8 July 2022]
• [16] F. Rosas, P. Mediano, “When and how to use Lempel-Ziv complexity”, Information Dynamics, 26 June 2019, [Online]. Available: https://information-dynamics.github.io/complexity/information/2019/06/26/lempel-ziv.html. [Accessed: 10 July 2022]
• [17] AW. Norman, HL. Henry, Hormones: Growth factors, Third Edition, Academic Press, 2015, pp. 363-379, Doi: 10.1016/B978-0-08-091906-5.000-3. [Accessed: 15 July 2022]
• [18] AC. Mitchell, PS. Briquez, JA. Hubbell, JR. Cochran, “Engineering growth factors for regenerative medicine applications”, Acta Biomater, Jan 2016, 1-12, Doi: 10.1016/j.actbio.2015.11.007. [Accessed: 16 July 2022]
• [19] X. Ren, M. Zhao, B. Lash, MM. Martino, Z. Julier, “Growth factor engineering strategies for regenerative medicine applications”, Frontiers in Bioengineering and Biotechnology journal, vol. 7, January 2020, Doi: 10.3389/fbioe.2019.00469. [Accessed: 16 July 2022]
• [20] İB. Çitçi, DA. Jafari, B. Kosova, Sağlık Bilimleri Alanında Akademik Çalışmalar-II: Nörotrofin ailesi, Gece Kitaplığı, vol. 2, pp. 333-349, June 2020. [Accessed: 31 July 2022]
• [21] S.Cohen, R. Levi-Montalcini, V. Hamburger, “A nerve growth-stimulating factor isolated from sarcom as 37 and 180”, Proc Natl Acad Sci USA, 40 (10): 1014-1018, Oct 1954; Doi: 10.1073/pnas.40.10.1014. [Accessed: 10 August 2022]
• [22] L. Aloe, “Rita Levi-Montalcini: the discovery of nerve growth factor and modern neurobiology”, Trends in Cell Biology, vol. 14(7), pp. 395-9, Jul 2004, Doi: 10.1016/j.tcb.2004.05.011. [Accessed: 10 August 2022]
• [23] M. Costandi, Nöroplastisite: Büyüme faktörleri ve hücre intiharı, Pan Yayıncılık, January 2019, pp. 49-51. [Accessed: 10 August 2022]
• [24] U. Suter, C. Angst, CL. Tien, CC. Drinkwater, RM. Lindsay, EM. Shooter, “NGF/BDNF chimeric proteins: analysis of neurotrophin specificity by homolog-scanning mutagenesis”, The Journal of Neuroscience: the official journal of the Society for Neuroscience, vol. 12, 1, pp. 306-318, Jan 1992, Doi: 10.1523/JNEUROSCI.12-01-00306.1992. [Accessed: 16 July 2022]
• [25] J. Leibrock, F. Lottspeich, A. Hohn, M. Hofer, B. Hengerer, P. Masiakowski, H. Thoenen, YA. Barde, “Molecular cloning and expression of brain-derived neurotrophic factor”, Nature, vol. 341, 6238, pp 149-152, Sep 1989, Doi: 10.1038/341149a0. [Accessed: 25 July 2022]
• [26] J. Langhnoja, L. Buch, P. Pillai, “Potential role of NGF, BDNF, and their receptors in oligodendrocytes differentiation from neural stem cell: an in vitro study”, Cell Biology International, vol. 45, issue 2, pp. 432-446, February 2021, Doi: 10.1002/cbin.11500. [Accessed: 25 July 2022]
• [27] PC. Maisonpierre, L. Belluscio, B. Friedman, RF. Alderson, SJ. Wiegand, ME. Furth, RM. Lindsay, GD. Yancopoulos, “NT-3, BDNF, and NGF in the developing rat nervous system: Parallel as well as reciprocal patterns of expression”, Neuron, vol. 5, issue 4, pp. 501-509, October 1990, Doi: 10.1016/0896-6273(90)90089-X. [Accessed: 25 July 2022]
• [28] C. Zeeh, “The Lempel Ziv algorithm”, uwaterloo.ca, January 16, 2003. [Online]. Available: https://ece.uwaterloo.ca/~ece611/LempelZiv.pdf. [Accessed: 10 July 2022]
• [29] T. Weissman, “Chapter 1. Lempel-Ziv compression”, web.stanford.edu, [Online]. Available: https://web.stanford.edu/class/ee376a/files/EE376C_lecture_LZ.pdf. [Accessed: 10 July 2022]
• [30] E. Roberts, “Dictionary-based compressors”, cs.stanford.edu, [Online]. Available: https://cs.stanford.edu/people/eroberts/courses/soco/projects/data-compression/lossless/lz78/index.htm. [Accessed: 10 July 2022]
• [31] ST. Brink, “Lempel-Ziv compression”, webdemo.inue.uni-stuttgart.de, [Online]. Available: https://webdemo.inue.uni-stuttgart.de/webdemos/03_theses/Lempel-Ziv-Compression/index.php?id=1. [Accessed: 10 July 2022]
• [32] G. Sharma, “Analysis of Huffman Coding and Lempel-Ziv-Welch (LZW) coding as data compression techniques”, International Journal of Scientific Research in Computer Science and Engineering, vol. 8, issue 1, pp. 37-44, Feb 2020. [Accessed: 12 July 2022]
• [33] YT. Tan, BA. Rosdi, “FPGA-based hardware accelerator for the prediction of protein secondary class via fuzzy K-nearest neighbors with Lempel-Ziv complexity-based distance measure”, Neurocomputing, vol. 148, pp. 409-419, January 2015, Doi: 10.1016/j.neucom.2014.06.001. [Accessed: 5 July 2022]
• [34] HB. Shen, J. Yang, KC. Chou, “Fuzzy KNN for predicting membrane protein types from pseudo-amino acid composition”, Journal of Theoretical Biology, 240, 9-13, June 2006, Doi: 10.1016/j.jtbi.2005.08.016. [Accessed: 20 July 2022]
• [35] Z. Bian, CM. Vong, PK. Wong, S. Wang, “Fuzzy KNN method with adaptive nearest neighbors”, IEEE Transactions on Cybernetics, vol. 52, no. 6, pp. 5380-5393, June 2022, Doi:10.1109/TCYB.2020.3031610. [Accessed: 20 July 2022]
• [36] LA. Zadeh, “Fuzzy sets”, Information and Control, vol. 8, issue 3, pp. 338-353, 1965, Doi: 10.1016/S0019-9958(65)90241-X. [Accessed 20 July 2022]
• [37] KC. Chou, CT. Zhang, “Review: prediction of protein structural classes”, Critical Reviews in Biochemistry and Molecular Biolog, 30, 275–349, 1995. [Accessed 25 November 2024]
• [38] S. Sinharay, “Jackknife Methods”, International Encyclopedia of Education (Third Edition), 229-231, 2010. [Accessed 25 November 2024]
• [39] R. Bondugula, O. Duzlevski, D. Xu, “Profiles and fuzzy k-nearest neighbor algorithm for protein secondary structure prediction”, In Proceedings of the 3rd Asia-Pacific Bioinformatics Conference pp. 85-94, 2005
• [40] G. Mirceva, A. Naumoski, A. Kulakov, “Classification of protein structures by using fuzzy KNN classifier and protein voxel-based descriptor”, Mathematical Modeling, 2(3), 116-118, 2018.