Cancer Genomics
project代做 – , 这是值得参考的project代写的题目
project Leader: Tim Cannings Project Helper: Miguel de Carvalho Expert: Adnan Akbar (Cambridge Cancer Genomics) Session: 1 (1 June 3 July)
Aim: To detect driver mutations using statistical model
What is a mutation? A mutation can be defined in several ways. But for this purpose, we are only looking into somatic single nucleotide variants (SNV) which can be defined as the substitution of the single nucleotide with genome. For example, in the human genome, coding sequence ACT changes to AGT indicating (C->G) as mutation.
What is a driver mutation? A mutation with selective advantage or positive selection over other mutations. Cancer is a disease of the genome caused by somatic mutations. In this regard, most of the mutations are passenger and of benign nature with only few driver mutations. Driver mutations are responsible for driving the cancer. In precision oncology, identifying these driver mutations accurately is the first main step.
How it can be done? From a clinical perspective, frequency of the mutations is the most important criteria i.e. if the region is being mutated several times then it shows that region is under selective pressure and is of significance nature. It is very unlikely to get random mutations at the same genomic region several times. But taking only frequency into account may result in many false positives as certain regions in human genome are more likely to get mutated based on different biological characteristics. Therefore, a better way is to estimate background mutation rate or mutability for that region and calculate significance score using binomial model.
Steps:
- An accurate model based on the frequency must be done a very large cohort of patients. A small dataset might induce bias in it. For this purpose, we will use two open source databases (TCGA+Genie) with over 80, 000 tumours sequenced.
- Mutability at every expected location will be estimated based on the context of mutation with respect to local context (neighbouring nucleotides) and global context (based on the gene).
- Frequency is calculated at every location from the above-mentioned datasets.
- A binomial model will be used to calculate the significance score
- Based on the significance score, a decision will be made if a mutation is driver or not
Useful Courses: Biomedical Data Science
References: This approach is based on the following two publications, we might give the students a freedom to estimate mutability based on their understanding. [1] Identifying recurrent mutations in cancer reveals widespread lineage diversity and mutational specificity (https://www.ncbi.nlm.nih.gov/pubmed/26619011) [2] Accelerating Discovery of Functional Mutant Alleles in Cancer (https://www.ncbi.nlm.nih.gov/pubmed/29247016)