=====Discovar Team 5 Update=====

Assembler takes only Illumina libraries: 

Ideally, PCR free, high coverage and insert size ~450bp

Had to use Fastuniq to remove duplicates:
	
16X coverage being inputted, somewhat low for what discovar wants as input


====Running Discovar====
Used fraction option limit input of files to only portion of reads

Needed to specify threads and maximum memory for the run as well

50% UCSF run showed much better results in N50 for contig and scaffold than 50% original data run and used less memory

Discovar performed much better with 2x250 reads vs 2x100 reads; more scaffolds of longer length

Want to use full data set when there is more RAM available

====BLAST results====

8th longest scaffold when nucleotide BLASTed matched a transcript variant of sea hare

metallothionein hit may be result of having cysteine rich scaffold

most common gene hit was ribosomal subunit 28S, which is a good sign because this gene is consistent across species

Want to run PRICE to find viral sequences that were found with blast

would create an assembly for the viral sequnce that was found and determine if sequence was integrated in the genome or are extranuclear 

Can map contigs to scaffolds to see if any contig has a different coverage than normal coverage 

====SSpace====

SSpace to do scaffolding after getting contigs

Scaffolds and contigs had been coming out identical sequences

used 50% UCSF contigs as input, using SW041 and SW042 files

run with old BWA 0.5, will re-run with bwa 0.7 version

SSpace merged a few scaffolds, but only added more Ns

no change in scaffold N50
only affected shorter contigs
number of scaffolds decreased by 20-50

probably due to not enough coverage of the assembly


====mitochondrion assembly====

Looked for contig that might have been mitochondrial (previous class iteration)
Took reads that mapped to the 2012 consensus sequence
Hiseq w018 and sw019 reads so far
mito size 14kb estiamte
used discovar sw018 data that mapped to 2012 seq-> coverage 60X


Want to use contigs built from read data rather than scaffold
start with one contig that maps well to mito  (use 12kb discovar 18+19 output)

mito genome does integrate into nuclear genome, over time mutates and changes sequence, results in lots of ambiguity in contig construction

seems like 12kb contig is entire mitochondria genome
2nd largest contig (3245bp) looks like might be missing part of the mito

look at ends of contigs and compare, try to join Ns together

sea hare is 14kb, usually doesnt include hypervariable region that is very difficult to assemble