Differences

This shows you the differences between two versions of the page.

--- lecture_notes:04-30-2010 [2010/05/02 22:05]
jstjohn
+++ lecture_notes:04-30-2010 [2010/05/05 21:05] (current)
jmagasin
@@ Line 303: / Line 303: @@
 Even just getting the right answer within a factor of 10 was
 encouraging.
+===== Second set of lecture notes =====
+From Jonathan Magasin.
+Today's lecture covered files generated by Kevin's map-colorspace5\\
+script which will be necessary for the homework assignment: ordering\\
+the Pog contigs.  Also covered: How to roughly estimate banana slug\\
+genome size.
+==== Homework assignment: assemble Pog ====
+From the newer newbler we have a different set of contigs, thirty-one\\
+of them.  We also have mate pairs from SOLiD mapped to newbler\\
+assembly 5.  The assignment is to order and orient them the contigs.\\
+Kevin has posted his solution in the assembly directory\\
+(assemblies/Pog/map-colorspace5).
+=== 'delete' and 'invert' files ===
+The trim9 files are output from Kevin's map-colorspace5 script.  Most\\
+of them are not for us.  The 'delete' files are for mate pairs where\\
+both mapped but not at the appropriate distance: closer than 350\\
+bases, or more than 5000 bases apart.  These files can be studied to\\
+find massive deletions.
+'invert' files are for mate pairs that were on opposite strands (at\\
+any distance) and are useful for studying inversions.  However\\
+inversions are not so useful because they will not be within a single\\
+contig.
+The 'between contigs' files are for when the reads mapped to different\\
+contigs.  Only cases with unique mappings are in these files.  fixme:\\
+What file(s) is this?
+=== trim9.out ===
+The trim9.out file is a summary of what happened during mapping.\\
+Kmers are colorspace kmers.  These kmers are the lengths after\\
+trimming off nine.  'uniquly mapped' means one for the R3 and one for\\
+the F3.  Note that some of the summary stats are bugs (the 'wrong\\
+range' line) that have since been fixed.  [Kevin reran the script of\\
+Friday.]
+Compared to earlier mapping, Kevin has increased the agressiveness.
+The deletions ('wrong range' line) appear to be from reads taken from\\
+clones with multiple adapters.
+Strand biases (algorithmic) were checked for in the forward and\\
+backward counts of mapped reads.
+Tonight Kevin [reran] the mapping script.  The new output will\\
+include error rates by position. Sol requested these be called\\
+mismatches rather than errors.  Kevin said they are estimates of\\
+sequencing error.  They are in colorspace, and exclude indels.  In\\
+reality they are differences between the reads and assemblies, not\\
+errors.  But for all practical purposes when there is a mismatch\\
+between a read and the assembly the error is in the read.)
+=== trim9.joins ===
+The data we'll use is not in trim9.out, rather it is in trim9.joins.\\
+trim9.joins is computed from trim9-cross.rdb (which has reads that\\
+crossed a contig boundary).
+Looking at the first line: How many reads support contig 22 following\\
+contig 3?  34K.  The contigs lengths are there to see if a contig is\\
+short enough that mate pairs might span them.  E.g. what orderings are\\
+consistent with the listed contig orders: 2->3->4, or maybe 3 is very\\
+short so we have a mate pair that jumps over 3.
+A minus sign appearing before a contig name means to take the\\
+reverse-complement of that contig.  (And contig3 followed by contig22\\
+is the same as -contig22 followed by -contig3.)
+The last number on each line indicates if the mate pair were the\\
+optimal length (optimal is the peak of length distribution).  That\\
+number is the expected length of the gap between the contigs.  It is\\
+very rough.
+The file is sorted most-counts to fewest.  The stuff at bottom of the\\
+file is noise.  Only about first fifty lines are helpful.
+Long mate pairs are very helpful for disambiguating, and wish we had\\
+them for //H.pylori//.
+=== Your task ===
+Take the trim9.joins file and try to order and orient the contigs,\\
+knowing that some of them are duplicated, and that there is a virus in\\
+the mix and there should be one large chromosome.

Banana Slug Genomics

User Tools

Site Tools

Differences

Page Tools