Banana Slug Genomics

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
===== April 21st, 2010 =====
==== Logistic related notes ====
  * There will be a guest lecture on Friday, April 23rd, 2010. Please do not be late!
  * By Friday, April 23rd, 2010, all published sources which contain information pertaining to Banana slug should be added to the [[:banana_slug_biology|Banana Slug Biology]] page. **Is this the right page?**
    * Expect 40-50 different resources (papers, books, phd theses)
    * The bibliography must be annotated, though annotations are not required this friday.
    * As you run assembly tool, keep the [[archive:computer_resources:assemblies|assemblies/]] wiki page updated!

==== Continuation of Newbler Assembler methods ====
  * By running Newbler using settings described last friday, Kevin obtained 43 contigs and ~2.4 megabases of DNA.
    * This measure is not bad compared to most other assemblers
=== Overview of directories ===
  * Newbler-clean
    * Not attempting to assemble data.
    * Used to remove contamination for H. Pylori which was sequenced on the same machine in the same run.
    * If you have a reference genome for your contaminants, you can try to clean them out computationally by mapping and removing matches.
    * Makefile Overview
      * Set up a new mapping
      * Maps reads from 454 Pog data to the contaminant genome
    * Interesting output: ReadStatus.txt
      * Contains mapping status of each read (Mapped, unmapped, partially mapped, too short)
      * Reads mapped to contaminants should be filtered out from 454 data and should not be sued for assembly.
    * Create a new SFF file (using sfffile utility) with unmapped data in file "no_Hyp.sff"
    * It may be beneficial to hang onto "too short" reads for use in assemblers which utilize shorter reads.
    * You can also use Megablast and NCBI's Taxonomy Report to identify contaminants.
  * Newbler-assembly2
    * Uses clean data (no H. Pylori) from newbler-clean1
    * One less small contig, mostly the same
  * Newbler    * 
    * Changed expected coverage to 60x-- This number is closer to the real coverage based on better estimates of genome size.
    * Still reported 41 contigs (or maybe one less?)
    * good news- All contigs map to reference genome
  * Map-colorspace3 (Map-colorspace directories begin indexing at 3 rather than 1)
    * The scripts ran in this directory were originally intended for finding inverseions from mate-pair reads.
    * New features have been added since then.
    * Now looks for reads between contigs
    * Tries to orient contigs
  * Newbler-partial3
    * attempts to use only partially mapped/unmapped reads
    * Plan is to later map contigs from this assembly to the full contigs from before (extend edges?)
    * Megablast, blastn, blat, find dna differences are four methods for mapping partially mapped reads onto contigs created by newbler.
      * Megablast and blastn showed very similar results
      * blat
        * output is a psl file
        * Nice results
        * Very sensitive but also very slow
        * Shows matches down to 14 bases in size
        * can handle intros and splice sites
        * May be too sensitive for some purposes
        * There are some parameters you can tweak which will alter the output.
          * Kevin thought it would be best to keep blat's default parameters and filter output.
        * A script was written to filter output and find the best matches.
        * For a resequencing project, take all contigs, map with blat, sort by start site and build a scaffold according to the order of the start sites of each contig.
      * Find dna differences
        * Shows differences between contigs and a reference genome in a human readable format
  * Newbler-assembly4
    * Adds partial3 contigs to full search as reads
    * Didn't seem to help much
    * The resulting output had fewer total bases and more contigs than originally - perhaps worse.
  * Newbler-assembly5
    * Utilized Sanger reads produced by David Bernick to join contigs
    * 45 Sanger reads total
      * Sanger reads were created by running pcr on conjectured contig join sites.
      * Two possible methods for guessing contig join sites
        * Use mate-pair data from solid run as evidence to create primers
        * David Bernick used know Crispr repeats. Some contigs showed the start of a repeat, while others showed the end
      * After adding Sanger reads, the final assembly went down to 31 contigs
      * In contig-length.rdb
        * Most reads have ~0.16 reads/base
          * The reads/base metric was utilized because computation was simpler than coverage.
        * Some reads had greater reads/base ~0.32 (qty. 3) and ~0.44 reads per base
          * Could 0.32 reads/base indicate two occurences of the contig in the genome?
          * Could 0.44 occur three times in the genome?
        * All 0.16 reads/base contigs occured a single time in the genome
        * 2/3 0.32 reads/base appeared twice in the genome.
        * 1/3 0.32 reads/base was a short contig and was within the expected deviation in reads/base from a sequencing run
        * 0.44 reads/base occured twice, not three times
    * Less contigs and more bases 
  * Map-colorspace5
    * map-colorspace has many parameters.
      * You can list all the parameters by issuing the command "map-colorspace --help"
    * Lots of output files using this command (~2 per contig). May prove problematic with assemblies that contain many contigs
    * Set length parameters using a histogram of reads mapped to contigs
    * merge_cross option merges all cross products into a single file rather than (number of contigs)^2 files
    * scaffold fasta & and color qualities are thrown out in this method
    * If F3 or R3 maps, the program aggresivly tries to map the paired read.
    * Output
      * Uniquely mapped reads mapped both F3 and R3 (Forward and Reverse) reads to the same contig
      * Multiply mapped- can map to multiple contigs
      * Wrong range, only F3 or R3 map
    * Summary File
      * Maps F3 and R3 from the end of one contig to the beginning of another
      * Shows that contigs are near, not touching.
      * short contigs may be skipped in paired end reads
    * QUESTION: Can we make a single connected graph using this data and all possible paths? -sometimes if data is particularly coherent. not guaranteed.
Banana Slug Genomics

User Tools

Site Tools

Page Tools
