Prostate Cancer

Prostate cancer or carcinoma of the prostate is a very complicated disease. Research on cancer indicates that prostate cancer is the second most diagnosed disease in the world. Moreover, splicing plays a crucial role in regulatory action on organisms’ transcriptomes. Around half of human genes have more than one splice variant. Spliced exons can seize several features that distinguish them from non-spliced ones, those features can be used to distinguish alternative from constitutive exons, transcription data set which has the alternative splicing events. In the present study, machine learning methods are used to extract the most relevant attributes in the gene transcription for prostate cancer data set. These can distinguish cancer from normal

cells or a cancer stage from another. Supervised machine learning methods are applied to learn from the training samples which have been taken from the data set, later tested it on the testing pool which samples, and create a mechanism to predict the class of the new samples. Cancer can be a described as a disease which is a result of accumulation somatic cellular aberrations. This accumulation can take place either in DNA or it can be the result of epigenetic changes.

Prostate cancer, a very common cancer, is considered as one of the most dangerous diseases and it has a very high death percentage. A research suggested that one in seven men will be diagnosed with prostate cancer in their lives CITATION Wal14 \l 1033 [1]. It is believed that there are three challenges in prostate cancer research. These challenges are risk assessment, differential diagnosis of aggressive and non-aggressive cancer, and developing new treatments for advanced disease. During the last decade, a lot of research has been carried out to understand the reasons, symptoms and treatment of Prostate cancer. Previously it was difficult to diagnose and treat prostate cancer, but recently tremendous progress has been made. Cancer cells have ability to grow uncontrollably. They can migrate from primary lesion and can create metastases in distant tissues. There is a lot of evidence available that suggests that alternative splicing, a process that allows the production of several mRNA variants from each gene, is responsible for the heterogeneity of the disease. The International Research Agency for Cancer has been working on estimating global cancer burden for last three

decades. A close analysis of the data gathered by IRAC suggests the risk of cancers differs from region to region. A simple justification for this analysis is that lifestyle, environment and exposure to known or unknown suspected risk factors vary from one place to other [2-6]. These factors create a challenge for prevention and treatment of the disease. Thus, personalized treatment for cancer is considered as a good option. For this purpose, it is necessary to adopt machine learning techniques. Regardless of challenges in machine learning, personalized treatment is a very good option CITATION Ten \l 1033 [3]. Research shows that cancer is a perpetually-evolving and multi-factorial disease. In many types of cancer, alternative splicing of genes is considered as a major post-transcriptional regulation mechanism. The application of next generation sequencing of RNA can provide us with a new technology which can be used to study transcriptomes. Moreover, it also gives us a systematic approach which can be used for qualitative study of alternative slicing. Similarly, it is quite evident that we need systematic tools because next generation sequencing is providing us a huge chunk of data [4-5]. If clinical diagnostics of cancer is managed properly, it could be useful in two ways. Firstly, accurate and comprehensive measurements of endpoints can be obtained. Secondly, more sophisticated analytical tools can provide us with more meaningful reflections of cellular states. Until now, few analytical tools have shown positive results regarding their ability to recognize tumor subtypes, predict clinical progression, and suggest different treatments CITATION MaS11 \l 1033 [4]. High-throughput mRNA sequencing have been conducted using various experiments and presented new algorithms which can be run onto the Cufflinks which is an open source software [7]. These experiments yielded some interesting results. It was found out that even in well-known model of muscle development; Cufflinks can illuminate substantial complexity and regularity. Moreover, it can also improve transcriptome based genome annotation [7]. Moreover, heterogeneity of genetic alterations in prostate cancer was investigated to find the complexity of the disease. It was suggested that genome-wide scans must be used while looking for nucleotide polymorphism and new technologies such as microarrays can be used for identification of new target genes which are involved in prostate cancer [6]. Global RNA sequencing was investigated in order to identify biomarkers of the recurrence. The global transcriptome analysis of formalin-fixed prostate could also be useful in identifying the biomarkers of the disease recurrence. The experiments not only predicted biochemical recurrence but also identified difference in gene expressions. In short, their research defines a biomarker panel that has potential to improve the clinical management of prostate cancer [8]. As mentioned previously, gene expression microarray can be used for the identification of new target genes involved in prostate cancer. This technology has multiple advantages such as, huge number of genes could be studies simultaneously [6-9]. Breen et al. (2006) have reviewed the procedure of applying microarrays in different areas of cancer research. Their research was focused on biomarkers, therapy, and improved diagnostics. Further advancement in this field can help in recognizing causes of different types of cancer, progression of cancer, treatment and clinical translation. Researchers have found out that although this technique is very promising, it is necessary to have knowledge of established techniques such as real time polymerase chain reaction (PCR). The only challenge for researchers in this field is to carefully analyze, design and perform microarray experiments [9]. To some extent, the work of H. Edgrenis (2012) linked with the work of Breen et al. (2006) Henrik has emphasized on using the microarrays technique which was also suggested by Breen et al [9-10]. However, the main focus of Henrik was on the identification of genome alterations in breast cancer. The work is divided into three parts. In the first part, the process of Gene Identification by Nonsense Inhibition method was investigated. Gini is a useful method in order to identify potential tumor presser genes. The first part of his work gave complicated results. In second part the work, a bioinformatics method was developed. This method was used for highly specific fusion gene identification. This phase of the work showed that fusion genes are more prevalent in breast cancer. Moreover, few more biological characteristics of fusion genes were also identified. The final phase of study was about using aCGH to characterize the size distribution of gene ERBB2 amplicon. The results of the study suggested that the cancer cells are dependent on a lot of genes other than the primary driver oncogene [10]. Gael P. Alamancos et al. (2013) have discussed methods to study splicing from high-throughput RNA sequencing data. They have debated that although it is practical to study splicing from high-throughput RNA sequencing data but the complexity of the information make it a very tough task. So, Gael P.

Alamancos and his team gave an overview of the available methods to study splicing. This overview can be used by newcomers who need to decide which tool is appropriate for them [14]. Tamara Steijgar et al. have also focused on RNA-seq. They have made an assessment of transcript reconstruction methods for RNA-seq. They evaluated 25 protocol variants of 14 independent computational methods for exon identification, transcript reconstruction, and expression level qualification from RNA-seq data. They found out very interesting results. Their results showed that most of the algorithms are able to find out discrete transcript components but complete assembly of isoform structure is still a challenge. They concluded that for the analysis of current generation RNA-seq data, the complexity of higher eukaryotic genomes creates a lot of limitations on splice product discrimination and transcript recall [15]. Detailed protein information can be obtained using ExPASy. The ExPASy is an online proteomics server which provides in-depth protein knowledge and information. There are specific analytical tools available which can be used for tasks related to proteomics. In short, The ExPASy is the best platform to get information regarding proteomics [11]. Rasoul and

David. (1990) have discussed effectiveness of androgen inhibitors in recurrent disease. Extensive experiments were performed to analyze the different aspects of the effectiveness of androgen inhibitors. A thorough study of results showed that heterogeneity in signaling pathways could contribute to the failure of cancer treatment [12]. It is evident that there are huge differences between different patients of prostate cancers. There is no specific mechanism available to understand these disparities. Moreover, a comprehensive landscape of the transcriptome profiles of 14 primary prostate cancers, have been conducted. Researchers did an analysis on correlation between genes and long ncRNA which revealed that long ncRNA might have functions other than transcriptional regulation. The study also revealed new insights into the pathogenesis of prostate cancer in the Chinese population [13].