dindels

Dindel is a computational method for calling small insertions and deletions (indels) from short-read sequencing data. It uses a Bayesian framework to evaluate the evidence for indels at candidate genomic positions and to assign posterior probabilities to possible indel genotypes. The approach accounts for sequencing error and alignment uncertainty, and it includes a local realignment step around potential indels to improve accuracy.

The workflow typically begins with aligned reads and a reference genome. For each candidate site, Dindel computes the likelihood of the observed reads under competing hypotheses (reference allele versus indel alleles) and combines these across reads to estimate genotype posteriors. The output comprises a list of indel calls along with confidence scores, often formatted in a VCF-like structure, enabling downstream filtering and interpretation.

Dindel is designed to detect small insertions and deletions, usually spanning a few base pairs to tens of base pairs. It has been used for germline indel discovery and, in some studies, for somatic analyses with matched normal samples. The method was influential in the development of more recent indel-calling approaches and remains cited for its explicit probabilistic treatment of indel evidence and alignment uncertainty. Limitations include reduced performance for longer or more complex indels, difficulties in highly repetitive regions, and dependence on sequencing depth and read quality.

See also: indels, indel realignment, short-read sequencing, Bayesian inference, variant calling.