Table 1.

RNAIndel features

		Feature identifiera	Feature description	Selection statusb
1	Sequence/Alignment	repeat	Count of repeat unit including homopolymers and STRs in indel flanking region	s, m
2		lc (linguistic complexity)	Diversity of k-mers in flanking 50-bp region
3		local_lc	Diversity of k-mers in flanking 6-bp region	s, m
4		gc	GC content in flanking 50-bp region
5		local_gc	GC content in flanking 6-bp region
6		strength	DNA pair-bond strength of 2-mers in flanking 50-bp region	m
7		local_strength	DNA pair-bond strength of 2-mers in flanking 6-bp region	s
8		dissimilarity**	Edit distance between indel and flanking sequences	m
9		indel_complexity	Mismatches around the indel measured by edit distance	s
10		indel_size**	Length of inserted or deleted nucleotides	m
11		is_ins	True for insertions	m
12		is_at_ins*	True for ‘A’ or ‘T’ insertions	s
13		is_at_del*	True for ‘A’ or ‘T’ deletions
14		is_gc_ins*	True for ‘G’ or ‘C’ insertions
15		is_gc_del*	True for ‘G’ or ‘C’ deletions	s
16		ref_count	Count of RNA-Seq reads representing the reference allele	s, m
17		alt_count	Count of RNA-Seq reads representing the indel allele	s, m
18		is_bidirectional	True if an indel is supported by forward and reverse reads	s
19		is_uniq_mapped	True if an indel is supported by uniquely mapped reads	s, m
20		is_near_exon_boundary	True if an indel is within exon but on the exon boundary	s, m
21		equivalence exists	True if alternative indel alignments are observe	s, m
22		is_multiallelic	True if multiple indels are observed at the locus	s, m
23	Transcript/Protein	is_inframe**	True if an indel is in-frame
24		is_splice	True if an indel is in an intronic region within 10-bp to exon
25		is_truncating	True if an indel causes frame-shift, or stop gain, or destroys splice motif
26		is_in_cdd**	True if an indel is located in conserved domain
27		indel_location	Relative indel location in coding region	s
28		is_nmd_insensitive	True if nonsense-mediate decay insensitive
29		indels_per_gene	Number of indels detected in the gene in the sample
30		cds_length	Length of the coding region	s
31	DB	is_on_db	True if indel is present in the default germline database	s, m

Note: A total of 31 features related to sequence/alignment, biological effect on transcription and protein coding, and match to germline variant database are examined.

^aFeatures marked with * were used only for training of single-nucleotide indel model while those marked with ** were used only for training of multi-nucleotide indel model.

^bFeatures selected by the single-nucleotide or the multi-nucleotide model are marked as s and m, respectively.