Short read DNA sequencing technologies from Illumina have made sequencing a human genome significantly more affordable, greatly accelerating studies of biological function and the association of genetic variants to disease. These technologies are frequently used to detect small genetic variants such as single nucleotide variants (SNVs) using a reference genome. However, short read sequencing technologies have several limitations. First, the human genome is diploid and short reads contain limited information for assembling haplotypes, or the sequences of alleles on homologous chromosomes. Moreover, there is significant input DNA required, which poses challenges for analyzing single cells. Further, there is limited ability to detect genetic variants inside long duplicated sequences that occur in the genome. As a result, there has been widespread development of novel methods to overcome these deficiencies using short reads. These include clone based sequencing, linked read sequencing, and proximity ligation sequencing, as well as various single cell sequencing methods. There are also entirely new sequencing technologies from Pacific Biosciences and Oxford Nanopore Technologies that produce significantly longer reads. While these emerging methods and technologies demonstrate improvements compared to short reads, they also have properties and error modalities that pose unique computational challenges. Moreover, there is a shortage of bioinformatics methods for accurate small variant detection and haplotype assembly using these approaches compared to short reads. This dissertation aims to address this problem with the introduction of several new algorithms for highly accurate haplotype assembly and SNV calling. First, it introduces HapCUT2, an algorithm that can rapidly assemble haplotypes using a broad range of sequencing technologies. Second, it introduces an algorithm for variant calling and haplotyping using SISSOR, a recently introduced microfluidics based technology for sequencing single cells. Finally, it introduces Longshot, an algorithm for detecting and phasing SNVs using error-prone long read technologies. In each case, the algorithms are benchmarked using multiple real whole-genome sequencing datasets and are found to be highly accurate. The methods introduced in this dissertation contribute to the goal of sequencing diploid genomes accurately and completely for a broad range of scientific and clinical purposes.