Bioinformatics Concepts and Tools

This guide explains the key bioinformatics concepts implemented in Mycelia and provides guidance on when to use each tool. It's designed for graduate-level researchers who need to understand both the biological context and computational approaches.

Workflow Overview
Data Acquisition
Quality Control
Sequence Analysis
Genome Assembly
Gene Annotation
Comparative Genomics
Phylogenetics
Tool Selection Guide

Workflow Overview

Bioinformatics analyses typically follow a structured workflow. Understanding this flow helps you choose the right tools and interpret results correctly.

Standard Genomic Analysis Pipeline

Raw Data → Quality Control → Feature Extraction → Assembly → Validation → Annotation → Comparative Analysis

Decision Points

At each stage, you face key decisions:

Data Quality: Is my data sufficient for analysis?
Method Selection: Which algorithm fits my data type and research question?
Parameter Tuning: How do I optimize for my specific dataset?
Result Interpretation: What do the numbers mean biologically?

Data Acquisition

Concept: Data Sources and Types

Biological Context: Genomic data comes from various sources, each with different characteristics and limitations.

Data Types

Sequencing Data

Illumina: Short reads (150-300 bp), high accuracy, paired-end
PacBio/HiFi: Long reads (10-25 kb), high accuracy, single-molecule
Oxford Nanopore: Ultra-long reads (50+ kb), moderate accuracy, real-time

Reference Data

RefSeq: Manually curated, high-quality reference sequences
GenBank: Community-submitted sequences, variable quality
SRA: Raw sequencing data from published studies

When to Use Each

Data Type	Use Case	Advantages	Limitations
Illumina	SNP calling, RNA-seq, metagenomics	High accuracy, cost-effective	Short reads, repetitive regions
PacBio HiFi	Genome assembly, structural variants	Long + accurate, span repeats	Higher cost, lower throughput
Nanopore	Real-time analysis, ultra-long reads	Longest reads, portable	Higher error rates

Mycelia Functions

# Download reference genomes
Mycelia.download_genome_by_accession(accession="NC_001422.1")

# Download complete assemblies
ncbi_genome_download_accession(accession="GCF_000819615.1")

# Simulate test data
simulate_random_genome(length=10000, gc_content=0.45)
simulate_hifi_reads(genome, coverage=20, error_rate=0.001)

Quality Control

Concept: Data Quality Assessment

Biological Context: Sequencing is not perfect. Raw data contains errors, biases, and artifacts that must be identified and corrected before analysis.

Quality Metrics

Sequence Quality

Phred Scores: Probability of base-calling error (Q20 = 1% error, Q30 = 0.1% error)
GC Content: Should match expected organism profile
Length Distribution: Indicates fragmentation or size selection

Coverage Metrics

Read Depth: Number of reads covering each position
Uniformity: Even coverage across genome
Duplication Rate: PCR duplicates inflate coverage

Quality Issues

Common Problems

Adapter Contamination: Sequencing adapters in reads
Low Quality Ends: Error-prone read ends
Overrepresented Sequences: PCR bias or contamination
Length Bias: Preferential sequencing of certain sizes

Solutions

Trimming: Remove low-quality bases and adapters
Filtering: Discard low-quality reads
Normalization: Adjust for coverage bias
Deduplication: Remove PCR duplicates

When to Apply QC

Analysis Type	QC Priority	Key Metrics
Genome Assembly	Critical	Coverage uniformity, read length
Variant Calling	High	Base quality, mapping quality
RNA-seq	High	rRNA contamination, 3' bias
Metagenomics	Medium	Host contamination, diversity

Mycelia Functions

# Analyze read quality
analyze_fastq_quality("reads.fastq")

# Calculate sequence statistics
calculate_sequence_stats(sequences)

# Quality visualization
plot_quality_distribution(quality_scores)

Sequence Analysis

Concept: K-mer Analysis

Biological Context: K-mers are subsequences of length k. They capture local sequence composition and are fundamental to many bioinformatics algorithms.

K-mer Theory

Mathematical Foundation

K-mer Space: 4^k possible DNA k-mers
Frequency Spectrum: Distribution of k-mer frequencies
Coverage Estimation: Genome size estimation from k-mer frequencies

Biological Interpretation

Repetitive Elements: High-frequency k-mers indicate repeats
Sequencing Errors: Low-frequency k-mers often represent errors
Genome Complexity: Unique k-mers indicate complex regions

K-mer Applications

Genome Assembly

Error Correction: Remove k-mers with low frequency
Overlap Detection: Find shared k-mers between reads
Graph Construction: Build de Bruijn graphs

Quality Assessment

Coverage Estimation: Estimate genome size and coverage
Contamination Detection: Identify foreign DNA
Ploidy Estimation: Detect polyploidy from k-mer frequencies

Parameter Selection

Choosing K

Small K (k=15-21): Sensitive to errors, good for error correction
Medium K (k=25-31): Balanced sensitivity/specificity
Large K (k=35-51): Specific but may miss overlaps

Memory Considerations

Dense Matrices: Store all possible k-mers (memory intensive)
Sparse Matrices: Store only observed k-mers (memory efficient)
Counting Algorithms: Bloom filters, Count-Min sketch

When to Use K-mer Analysis

Application	K-mer Size	Matrix Type	Use Case
Error Correction	15-21	Dense	Remove sequencing errors
Assembly	25-31	Sparse	Genome assembly
Repeat Detection	31-51	Sparse	Identify repetitive elements
Contamination	21-31	Sparse	Detect foreign DNA

Mycelia Functions

# Count k-mers
count_kmers("reads.fastq", k=21)

# Dense vs sparse counting
fasta_list_to_dense_kmer_counts(files, k=21)
fasta_list_to_sparse_kmer_counts(files, k=21)

# K-mer spectrum analysis
kmer_frequency_spectrum(kmer_counts)

Genome Assembly

Concept: Reconstructing Genomes

Biological Context: Sequencing breaks genomes into fragments. Assembly reconstructs the original genome by finding overlaps between fragments.

Assembly Algorithms

Overlap-Layout-Consensus (OLC)

Overlap: Find overlaps between all reads
Layout: Order reads based on overlaps
Consensus: Generate consensus sequence
Best For: Long reads (PacBio, Nanopore)

De Bruijn Graph

K-mer Decomposition: Break reads into k-mers
Graph Construction: Connect overlapping k-mers
Path Finding: Find Eulerian paths through graph
Best For: Short reads (Illumina)

String Graph

Overlap Graph: Nodes are reads, edges are overlaps
Transitivity Reduction: Remove redundant edges
Contig Construction: Find paths through graph
Best For: Long reads with high accuracy

Assembly Challenges

Repetitive Sequences

Problem: Identical sequences confuse assembly
Solution: Long reads that span repeats
Tools: Repeat-aware assemblers, scaffolding

Heterozygosity

Problem: Diploid genomes have two alleles
Solution: Haplotype-aware assembly
Tools: Trio binning, Hi-C scaffolding

Polyploidy

Problem: Multiple copies of chromosomes
Solution: Specialized polyploid assemblers
Tools: Ploidy-aware algorithms

Assembly Quality Metrics

Contiguity

N50: Length where 50% of assembly is in contigs of this length or longer
L50: Number of contigs containing 50% of assembly
Contig Count: Total number of contigs (fewer is better)

Completeness

Genome Coverage: Percentage of genome assembled
Gene Completeness: Percentage of expected genes found
BUSCO Scores: Conserved gene completeness

Accuracy

Base Accuracy: Percentage of correct bases
Structural Accuracy: Correct arrangement of sequences
Validation: Comparison with reference genome

Gene Annotation

Concept: Identifying Functional Elements

Biological Context: Raw genome sequences are meaningless without annotation. Gene annotation identifies protein-coding genes, regulatory elements, and other functional sequences.

Annotation Types

Structural Annotation

Gene Prediction: Identify protein-coding genes
Exon-Intron Structure: Define gene boundaries
Promoter Prediction: Identify regulatory sequences
Repeat Annotation: Classify repetitive elements

Functional Annotation

Protein Function: Predict what proteins do
Pathway Mapping: Assign genes to metabolic pathways
GO Terms: Gene Ontology functional classification
Domain Annotation: Identify protein domains

Gene Prediction Methods

Ab Initio Prediction

Approach: Use sequence signals (start codons, splice sites)
Advantages: No external data required
Limitations: Lower accuracy, misses non-canonical genes
Tools: Prodigal, Pyrodigal, Augustus, GeneMark

Homology-Based Prediction

Approach: Compare to known genes in databases
Advantages: Higher accuracy for conserved genes
Limitations: Misses novel genes, depends on database quality
Tools: BLAST, DIAMOND, MMseqs2, MetaEuk

RNA-seq Guided Prediction

Approach: Use transcriptome data to guide prediction
Advantages: High accuracy, identifies expressed genes
Limitations: Requires RNA-seq data, misses lowly expressed genes
Tools: StringTie, Cufflinks, BRAKER

Annotation Challenges

Eukaryotic Complexity

Alternative Splicing: Multiple transcripts per gene
Pseudogenes: Non-functional gene copies
Non-coding RNAs: Functional RNAs that don't code for proteins

Prokaryotic Specifics

Operons: Polycistronic transcripts
Overlapping Genes: Genes sharing sequence
Horizontal Gene Transfer: Genes from other species

Quality Assessment

Annotation Completeness

Gene Density: Number of genes per kb
Protein Completeness: Percentage of complete proteins
Functional Coverage: Percentage of genes with functional annotation

Annotation Accuracy

Validation: Comparison with experimental data
Consistency: Agreement between different methods
Benchmarking: Comparison with reference annotations

When to Use Different Approaches

Genome Type	Method	Tools	Considerations
Bacterial	Ab initio	Prodigal / Pyrodigal	Simple gene structure
Viral	Specialized	Prodigal-gv	Viral coding patterns
Fungal	Hybrid	Augustus + BLAST	Moderate complexity
Eukaryotic metagenome	Homology-guided	MetaEuk	Fragmented contigs
Plant/Animal	RNA-seq guided	BRAKER	High complexity, alternative splicing

Synteny: Conserved gene order between species
Orthology: Corresponding genes in different species
Whole Genome Alignment: Align entire genomes
Structural Variation: Differences in genome structure