Bio Informatics (AL-803 (B)) - Important Questions

Explain gene finding strategies: ab initio methods and evidence-based (extrinsic) approaches. Compare their principles, advantages and limitations, and describe scenarios where a hybrid approach is preferred.

Describe the probabilistic models used in ab initio gene prediction with emphasis on Hidden Markov Models (HMMs). Explain how states, transitions and emission probabilities are used to model gene structure (exons, introns, intergenic regions) and how the Viterbi algorithm is applied for gene prediction.

List and compare commonly used gene prediction tools such as GENSCAN, AUGUSTUS, GeneMark, Glimmer and SNAP. For each tool indicate the primary algorithmic approach (e.g., HMM, heuristic, SVM), typical input data, and typical applications.

How are splice sites and promoter regions predicted computationally? Describe common sequence signals, position weight matrices (PWMs) or motif models used, scoring schemes, and how these predictions are integrated into gene prediction pipelines.

Outline a typical computational pipeline for mining gene expression data from raw sequencing counts to differential expression results. Include steps for quality control, filtering, normalization, statistical testing, multiple testing correction, visualization and mention commonly used tools (e.g., FastQC, DESeq2, edgeR, limma).

Explain methods for normalization of gene expression data. Compare RPKM/FPKM and TPM for within-sample normalization and discuss count-based normalization approaches such as DESeq2 size factors and edgeR TMM for between-sample normalization.

Describe clustering techniques used in gene expression analysis, specifically hierarchical clustering and k-means clustering. Explain choice of distance measures (e.g., Euclidean, Pearson correlation), linkage criteria for hierarchical clustering, and how results are interpreted via heatmaps and dendrograms.

Discuss functional enrichment analysis after obtaining a set of differentially expressed genes. Explain Gene Ontology (GO) enrichment and pathway analysis approaches and describe the necessity of multiple testing correction (for example Benjamini–Hochberg) when reporting adjusted $p$-values and thresholds such as adjusted $p < 0.05$ and fold change criteria (e.g., $\log_2$ fold change).