The VCF files contain small indel calls for the CEU and YRI trios. The calls were made from MAQ-aligned Broad-recalibrated Illumina BAM files. 454 and SOLID data were not considered in the analysis. The calls were made as follows: 1. Extract all indels in MAQ alignments from the BAM file. These are the candidate indels. 2. Generate genotype likelihoods using Dindel for every candidate indel 3. Assuming Mendelian inheritance, calculate the posterior probability for each candidate indel in all three trio members jointly. Thus, the VCF files only contain indels that are consistent with Mendelian inheritance. Independent analysis of the trio members indicated that there is signficant number of cases where the trio-child shows evidence for an indel, but none of the parents does; these cases are not included in the VCF file. Such cases are likely explained by reduced sensitivity for detecting indels as compared to SNPs, mapping errors, and some may be a result of de novo mutations. We hope to address these issues in future releases by using more sensitive mapping and indel calling algorithms. NOTE The naming convention for the indels deviates from the specification V3.3 of the VCF format. The naming convention is explained in the header of the VCF file, and is consistent with the one use for the Dindel indel call set for pilot 1. NOTE Only the lines in the VCF file where the filter column is "0" should be considered as calls. If the filter column is not "0", one of the filters described in the header of the VCF file apply and the indel should not be considered as a call. NOTE As described above, all reported calls are consistent with Mendelian inheritance by construction. DINDEL A test version of Dindel can be downloaded from http://sites.google.com/site/keesalbers/soft/dindel If you have any questions, please contact me (Kees) at caa@sanger.ac.uk Kees Albers Gerton Lunter Richard Durbin