SOAPdenovo issue

[szimmerman]

Hi everyone,

I am getting some strange results after using SOAPdenovo. First I run prepare, then I run map and scaff like usual. My problem is, that I load 6886 contigs into the scaff, but then all the sudden later in my results for scaff it says there are a total of 13772 contigs. Does anyone have an idea what might have happened? Below I will copy and paste the output I got from scaff and put in bold the discrepancy I am talking about. Thanks!

Scaff
********************

Parameters: scaff -g ICH55

ICH55.Arc: no such file or empty file!

There are 1 grad(s), 277742070 read(s), max read len 101.
Kmer size: 55
There are 13772 edge(s) in edge file.
Mask contigs with coverage lower than 0.3 or higher than 6.0, and strict length 0.
Average contig coverage is 3, 0 contig(s) masked.
Mask contigs shorter than 57, 134 contig(s) masked.
0 arc(s) loaded, average weight is 0.
6886 contig(s) loaded.
Done loading updated edges.
Time spent on loading updated edges: 2s.

*****************************************************
Start to load paired-end reads information.

For insert size: 270
Total PE links 102415205
Normal PE links on same contig 101347414
Incorrect oriented PE links 712334
PE links of too small insert size 336470
PE links of too large insert size 3060
Correct PE links 9412
Accumulated connections 1966
Use contigs longer than 270 to estimate insert size:
PE links 101326878
Average insert size 95
SD 175
983 new connections.

All paired-end reads information loaded.
Time spent on loading paired-end reads information: 293s.

*****************************************************
Start to construct scaffolds.

***************************
For insert size: 270
Total PE links 982
PE links to masked contigs 6
On same scaffold PE links 0
Cutoff of PE links to make a reliable connection: 3
Active connections 1952
Weak connections 850
Weak ratio 43.5%
7 circles removed.
Start to remove transitive connection.
Total contigs 13772
Masked contigs 162
Remained contigs 13610
None-outgoing-connection contigs 12655 (92.983101%)
Single-outgoing-connection contigs 894
Multi-outgoing-connection contigs 17
Cycle 1
Two-outgoing-connection contigs 44
Potential transitive connections 1
Transitive connections 0
Transitive ratio 0.0%
Start to linearize sub-graph.
Picked sub-graphs 58
Connection-conflict 0
Significant overlapping 55
Eligible 0
Bubble structures 0
Mask repeats:
Puzzles 43
Masked contigs 36
Start to remove transitive connection.
Total contigs 13772
Masked contigs 234
Remained contigs 13538
None-outgoing-connection contigs 12686 (93.706604%)
Single-outgoing-connection contigs 845
Multi-outgoing-connection contigs 1
Cycle 1
Two-outgoing-connection contigs 6
Potential transitive connections 1
Transitive connections 0
Transitive ratio 0.0%
Start to linearize sub-graph.
Picked sub-graphs 6
Connection-conflict 0
Significant overlapping 6
Eligible 0
Bubble structures 0
Non-strict linearization.
Start to linearize sub-graph.
Picked sub-graphs 5
Connection-conflict 0
Significant overlapping 4
Eligible 0
Bubble structures 0
Start to mask puzzles.
Masked contigs 3
Remained puzzles 0
Freezing done.

Recover contigs.
Total recovered contigs 0
Single-route cases 0
Multi-route cases 0

All links loaded.
Time spent on constructing scaffolds: 0s.

The final rank

*******************************
Scaffold number 345
In-scaffold contig number 6832
Total scaffold length 17430856
Average scaffold length 50524
Filled gap number 0
Longest scaffold 647306
Scaffold and singleton number 6408
Scaffold and singleton length 61984163
Average length 9672
N50 53501
N90 4087
Weak points 0

*******************************

Done with 345 scaffolds, 0 gaps finished, 424 gaps overall.

Overall time spent on constructing scaffolds: 5m.

[jebowers]

Soap denovo counts the result of contigs >100bp as a quality assessment for the contig step. It also makes all contigs down to a size of the kmer+1. It keeps all the contigs from k+1 to 100 for the scaffolding stage later (options allow them to be masked) but they are going to be there.

The confusion is that scaffolding reports on the k+1 or larger contigs, while other parts of the program talk about 100bp+ contigs.

Generally lots of the small contigs are based on errors in sequencing or are at the boundary between 2 repeats segments. (repeat) (short contig) (repeat).