Supplementary: An Integrative Approach to Predicting the Functional Effects of Non-Coding and Coding Sequence Variation

A. 46-Way Sequence Conservation: variants within highly conserved regions are likely to have more impact than those within high variability regions; therefore we used two measures of evolutionary conservation, namely PhastCons (Siepel et al., 2005) and PhyloP (Pollard et al., 2010) scores, obtained from the multiple sequence alignment (at the nucleotide level) of 46 vertebrate genomes to the human genome (Blanchette et al., 2004). In addition to these scores, we constructed ab initio hidden Markov models (HMMs) representing these alignments using the HMMER software package (version 3.1b1) and extracted the relative probabilities of each nucleotide at the corresponding position within the alignment. Following our previous work (Shihab et al., 2013b,a, 2014), we also included a measure of the magnitude of effect given the SNV (i.e. the log-odds ratio of observing both nucleotides). B. Histone Modifications (ChIP-Seq): we used ChIP-Seq peak calls for 14 histone modifications across 45 cell lines from ENCODE (The ENCODE Project Consortium, 2012). C. Transcription Factor Binding Sites (TFBS PeakSeq): we used PeakSeq (Rozowsky et al., 2009) peak calls for 119 transcription factors across 77 cell lines from ENCODE. D. Open Chromatin (DNase-Seq): we used DNase-Seq peak calls across 119 cell lines from ENCODE. E. 100-Way Sequence Conservation: we used similar features to those described in A. but now obtained from the multiple sequence alignment of 100 vertebrate genomes to the human genome. We considered both 100-way (E) and 46-way sequence conservation (A) to highlight any gain which could be made by including more species in the comparison. F. GC Content: we used a single measure for GC content calculated using a span of 5 nucleotide bases from the UCSC Genome Browser (Kent et al., 2002). G. Open Chromatin (FAIRE): we used Formaldehyde-Assisted Isolation of Regulatory Elements (FAIRE) peak calls across 119 cell lines from ENCODE. H. Transcription Factor Binding Sites (TFBS SPP): we used SPP peak calls Kharchenko et al. (2008) for 119 transcription factors across 77 cell lines from ENCODE. I. Genome Segmentation: we used 7 genome-segmentation states in 6 cell lines using a consensus merge of segmentations produced by the ChromHMM (Ernst and Kellis, 2010) and Segway software (Hoffman et al., 2012). J. Footprints: we used annotations describing DNA footprints across 41 cell types from ENCODE.