A scalable and memory-efficient algorithm for de novo transcriptome assembly of non-model organisms

BMC Genomics

Table 2 Comparisons of transcriptome assemblies of SOAPdenovo-Trans, Trans-ABySS and ASplice in model organisms over different values of k and k-mer coverage cutoff c

S. pom	SOAPdenovo-Trans					Trans-ABySS					ASplice
			total	unique	mem			total	unique	mem	splicing		total	unique	mem
k_c	locus	N50	hits	hits	(GB)	trans	N50	hits	hits	(GB)	graphs	N50	hits	hits	(GB)
25_10	3267	4455	7343	4230	10	21215	2854	28376	4400	3	5859	3032	5271	4650	9
25_20	3393	3795	7153	4341	10	14193	2880	19541	4284	3	5231	2723	4941	4579	9
25_50	3747	3025	8174	4513	10	8748	2317	10686	4370	3	5163	2210	5002	4580	9
31_10	3366	4164	8481	4342	10	20569	2977	33192	4323	4	5580	3005	5148	4611	9
31_20	3470	3590	7864	4418	10	13701	3045	22303	4158	4	5076	2625	4956	4565	9
31_50	3891	2788	8783	4576	10	7972	2421	10259	4284	4	5103	2088	5080	4620	9
A. tha	SOAPdenovo-Trans					Trans-ABySS					ASplice
			total	unique	mem			total	unique	mem	splicing		total	unique	mem
k_c	locus	N50	hits	hits	(GB)	trans	N50	hits	hits	(GB)	graphs	N50	hits	hits	(GB)
25_10	18980	1705	80644	21614	17	348069	460	383663	21715	7	103450	344	92605	21487	10
25_20	16327	1622	71684	20239	17	159952	772	139875	20213	7	71419	561	66645	20033	9
25_50	13384	1472	53209	17788	17	66058	948	70136	17544	7	43407	778	42396	17535	9
31_10	19604	1700	74854	20952	17	350165	578	525359	21422	7	92665	444	88408	21288	10
31_20	16882	1605	62748	19516	17	141642	948	181322	19838	7	58877	760	56869	19691	9
31_50	13660	1438	42763	16990	17	54189	1083	75863	17050	7	35448	973	34906	17030	9
D. mel	Trans-ABySS								ASplice
			total		unique		mem		splicing			total	unique		mem
k_c	trans	N50	hits		hits		(GB)		graphs		N50	hits	hits		(GB)
25_10	135048	1192	179499		12989		27		99930		728	51854	12322		11
25_20	83303	1693	102591		12848		27		60662		1328	35402	12245		10
25_50	47341	2082	58523		12453		27		36093		1874	26130	11921		9
31_10	113805	1547	225887		13025		30		77439		1203	45278	12453		11
31_20	70061	2029	124050		12787		30		45645		1952	28616	12259		10
31_50	41210	2296	64736		12337		30		32593		2085	24093	11861		9

The predicted units are locus for SOAPdenovo-Trans that is represented as a splicing graph containing nodes and edges, transcript (trans) for Trans-ABySS that is a linear concatenation of constituent nodes, and splicing graph for ASplice. For SOAPdenovo-Trans and ASplice, N50 denotes the N50 value of the length (in nucleotides) of the longest path in each splicing graph. For Trans-ABySS, N50 denotes the N50 value of the length of a predicted transcript, and only predicted transcripts of length at least 100 are retained. Total hits denotes the total number of hits from nucleotide BLAST search of nodes to the transcriptome of the same organism. Isoforms are considered to be the same gene. Only the top hit with E-value below 10⁻⁷ is considered. Hits from nodes within the same predicted unit to the same gene are counted only once. Unique hits denotes the number of unique hits to different genes. Mem (GB) denotes the physical memory requirement in gigabytes over all stages of each algorithm

ISSN: 1471-2164