Fig. 1
From: Efficient HLA imputation from sequential SNPs data by transformer
Architectures of HLARIMNT. a Overview of the model structure. b Embedding Layer. c Transformer Layer. d Classification Layer. HLARIMNT takes the input of each haplotype SNP genotype from pre-phased data represented as binary vectors, and outputs the genotype dosages of alleles for each HLA gene. input len: the number of SNPs to be input in Embedding Layer; chunk num: the number of chunks into with the SNPs are divided in Embedding Layer; chunk len: the number of SNPs in a single chunk; class num: the dimension of the output of Classification Layer
