Skip to content

usage-guide.md

Latest commit

 

History

History
56 lines (42 loc) · 2 KB

usage-guide.md

File metadata and controls

56 lines (42 loc) · 2 KB

Long-Read Alignment - Usage Guide

Overview

minimap2 is the standard aligner for long reads. It's fast, accurate, and handles the higher error rates of long-read data. Different presets optimize for ONT, PacBio, or RNA sequencing.

Prerequisites

conda install -c bioconda minimap2 samtools

Quick Start

Tell your AI agent what you want to do:

  • "Align my Nanopore reads to the reference genome"
  • "Map PacBio HiFi reads and generate a sorted BAM"

Example Prompts

Basic Alignment

"Align my ONT reads in reads.fastq.gz to reference.fa and output a sorted BAM"

"Map my PacBio HiFi reads to the human genome using minimap2"

With Read Groups

"Align reads with sample ID 'patient1' and add read group information for downstream GATK"

RNA/cDNA

"Align my Nanopore direct RNA reads to the transcriptome using the splice preset"

Performance

"Build a minimap2 index for the reference genome so future alignments are faster"

What the Agent Will Do

  1. Select the appropriate minimap2 preset based on your data type (map-ont, map-hifi, map-pb, splice)
  2. Run alignment with proper threading
  3. Pipe output through samtools for sorting and indexing
  4. Check alignment statistics with flagstat

Choosing the Right Preset

Data Type Preset
Nanopore (any) map-ont
PacBio HiFi (Q20+) map-hifi
PacBio CLR (older) map-pb
cDNA/direct RNA splice

Tips

  • Always use the correct preset for your data type - wrong preset will give poor alignments
  • Add read groups with -R '@RG\tID:run1\tSM:sample1' for downstream tools like GATK
  • Pre-build index with minimap2 -d ref.mmi ref.fa for multiple alignment runs
  • Check alignment rate with samtools flagstat - low rates indicate wrong preset or reference mismatch
  • Use PAF format (-x without -a) when you only need alignment coordinates, not full BAM

Resources