LncRNA-Disease association prediction based on nonnegative matrix factorization

Ahmet Toprak

Article Details

LncRNA-Disease association prediction based on nonnegative matrix factorization

Authors:

Ahmet Toprak

Publisher:

Nigde Omer Halisdemir University

Main Topics:

Makine Öğrenme (Diğer)

Journal:

Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi Vol.12 Issue 4, pp.1194-1199

Document Type:

Article

Publication Date:

10/2023

Document Language:

Turkish

Document ID (Org No.):

2001301224

pISSN:

2564-6605

Keywords:

Abstract:

 Many biological experiments have proven that lncRNA is related to various complex human diseases. Therefore, knowing the lncRNA-disease relationships not only facilitates the diagnosis, treatment and prognosis of the disease helps to understand the disease mechanism. However, determining the lncRNA-disease relationships through biological experiments is both costly and time-consuming. For this reason, many researchers have suggested calculational methods to forecast potential relationships between lncRNAs and diseases. In this study, we suggest a computational method named NMF to forecast possible lncRNAs, based on the assumption that functionally similar lncRNAs tend to associate with phenotypically similar diseases. This method integrates the lncRNA expression similarity network, the lncRNA cosine similarity network, the disease semantic similarity network, the disease cosine similarity network, and the known lncRNA-disease relationship network. To demonstrate the prediction accuracy of our method, we applied 5-fold cross-validation and leave-out cross-validation techniques and obtained ROC plots. AUC of 0.7837 for 5-fold cross-validation and 0.8551 AUC for leave-out cross-validation were calculated. The results show that the NMF method has reliable prediction performance.

 Many biological experiments have proven that lncRNA is related to various complex human diseases. Therefore, knowing the lncRNA-disease relationships not only facilitates the diagnosis, treatment and prognosis of the disease helps to understand the disease mechanism. However, determining the lncRNA-disease relationships through biological experiments is both costly and time-consuming. For this reason, many researchers have suggested calculational methods to forecast potential relationships between lncRNAs and diseases. In this study, we suggest a computational method named NMF to forecast possible lncRNAs, based on the assumption that functionally similar lncRNAs tend to associate with phenotypically similar diseases. This method integrates the lncRNA expression similarity network, the lncRNA cosine similarity network, the disease semantic similarity network, the disease cosine similarity network, and the known lncRNA-disease relationship network. To demonstrate the prediction accuracy of our method, we applied 5-fold cross-validation and leave-out cross-validation techniques and obtained ROC plots. AUC of 0.7837 for 5-fold cross-validation and 0.8551 AUC for leave-out cross-validation were calculated. The results show that the NMF method has reliable prediction performance.

 Many biological experiments have proven that lncRNA is related to various complex human diseases. Therefore, knowing the lncRNA-disease relationships not only facilitates the diagnosis, treatment and prognosis of the disease helps to understand the disease mechanism. However, determining the lncRNA-disease relationships through biological experiments is both costly and time-consuming. For this reason, many researchers have suggested calculational methods to forecast potential relationships between lncRNAs and diseases. In this study, we suggest a computational method named NMF to forecast possible lncRNAs, based on the assumption that functionally similar lncRNAs tend to associate with phenotypically similar diseases. This method integrates the lncRNA expression similarity network, the lncRNA cosine similarity network, the disease semantic similarity network, the disease cosine similarity network, and the known lncRNA-disease relationship network. To demonstrate the prediction accuracy of our method, we applied 5-fold cross-validation and leave-out cross-validation techniques and obtained ROC plots. AUC of 0.7837 for 5-fold cross-validation and 0.8551 AUC for leave-out cross-validation were calculated. The results show that the NMF method has reliable prediction performance.

 Yapılan birçok biyolojik deneylerde lncRNAnın çeşitli karmaşık insan hastalıklarının gelişimi ile yakından ilişkili olduğunu kanıtlanmıştır. Bu nedenle lncRNA ve hastalık arasındaki ilişkiyi bilmek sadece hastalık mekanizmasını anlamaya yardımcı olmakla kalmaz, aynı zamanda hastalığın teşhisini, tedavisini ve prognozunu da kolaylaştırır. Fakat lncRNA ile hastalık arasındaki ilişkiyi biyolojik deneyler yoluyla belirlemek hem maliyetli hem de çok zaman gerektiren bir süreçtir. Bu sebeple birçok araştırmacı potansiyel lncRNA-hastalık ilişkilerini tahmin etmek için hesaplamalı yöntemler geliştirmişlerdir. Bu çalışmada, fonksiyonel olarak benzer lncRNAların fenotipik olarak benzer hastalıklarla ilişki kurma eğiliminde olduğu varsayımına dayanarak, potansiyel lncRNAları tahmin etmek için NMF adı verilen bir hesaplama yöntemi öneriyoruz. Bu yöntem lncRNA ekspresyon benzerlik ağını, lncRNA kosinüs benzerlik ağını, hastalık semantik benzerlik ağını, hastalık kosinüs benzerlik ağını ve bilinen lncRNA-hastalık ilişkilendirme ağını entegre etmektedir. Yöntemimizin tahmin doğruluğunu göstermek için 5-katlı çapraz doğrulama ve birini dışarıda bırak çapraz doğrulama tekniklerini uyguladık ve ROC grafiklerini elde ettik. 5-katlı çapraz doğrulama için 0.7837 AUC değeri, birini dışarıda bırak çapraz doğrulama için 0.8551 AUC değeri hesaplandı. Sonuçlar NMF yönteminin güvenilir tahmin performansına sahip olduğunu göstermektedir.

Article Details

LncRNA-Disease association prediction based on nonnegative matrix factorization

Authors:

Ahmet Toprak

Publisher:

Nigde Omer Halisdemir University

Main Topics:

Makine Öğrenme (Diğer)

Journal:

Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi Vol.12 Issue 4, pp.1194-1199

Document Type:

Article

Publication Date:

10/2023

Document Language:

Turkish

Document ID (Org No.):

2001301224

pISSN:

2564-6605

lncRNA

disease

lncRNA-disease association

NMFlncRNA

hastalık

lncRNA-hastalık ilişkisi

NMF

Issues

Article Details

Authors: Ahmet Toprak

Publisher: Nigde Omer Halisdemir University

Main Topics: Makine Öğrenme (Diğer)

Journal: Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi Vol.12 Issue 4, pp.1194-1199

Document Type: Article

Publication Date: 10/2023

Document Language: Turkish

Document ID (Org No.): 2001301224

pISSN: 2564-6605

Issues

Authors:

Ahmet Toprak

Publisher:

Nigde Omer Halisdemir University

Main Topics:

Makine Öğrenme (Diğer)

Journal:

Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi Vol.12 Issue 4, pp.1194-1199

Document Type:

Article

Publication Date:

10/2023

Document Language:

Turkish

Document ID (Org No.):

2001301224

pISSN:

2564-6605