interpreting assembly contiguity statistics

Question

[goatsrunfaster]

I have the following assembly contiguity statistics and am just trying to get some inference as to which assembly is the best. My understanding is that the higher the N50 the better, but just wondering how someone with a better-trained eye would interpret the statistics below (given that some of the N50s aren't that far off):

Assembly 1:

n n:200 L50 min N80 N50 N20 max sum

4543321 422753 51496 200 1369 3815 8283 59877 718.7e6

Assembly 2:

n n:200 L50 min N80 N50 N20 max sum

3803625 420922 49315 200 1442 4079 8928 61614 740.8e6

Assembly 3:

n n:200 L50 min N80 N50 N20 max sum

3291053 420852 47950 200 1512 4288 9425 61630 758.9e6

Assembly 4:

n n:200 L50 min N80 N50 N20 max sum

2878387 428798 47104 200 1564 4448 9805 74286 775.5e6

Assembly 5:

n n:200 L50 min N80 N50 N20 max sum

2569027 441260 47057 200 1588 4552 9992 73154 790.9e6

Assembly 6:

n n:200 L50 min N80 N50 N20 max sum

2330974 458008 47663 200 1586 4582 10040 66524

Answer 1

[lieven.sterck]

ABySS assemblies I assume? ;)

yes, going with N50, combined with total length (should be as close as possible to expected genome size) is usually a good criterium. If that does not give a conclusion, add in total amount of sequences, (the lesser the better)