GalaxyFilterCommonSnps

Protocol for Filtering common SNPs from a set of alignments

Galaxy supports set operations on single columns. Thus, I build an index column for each sample formated as "chr:pos:ref:alt", which I refer to as the indexAlt.

Tools used

• Text Manipulation
  • Compute
  • Cut
  • Concatenate Datasets tail-to-head
  • Filter data on any column
• Join, Subtract and Group
  • Group data by a column
  • Compare two Datasets

Step by Step

1. For each sample
   1. Creating the BAM files (usually with BWA + GATK realigner)
   2. Create VCF of variant SNPs (mpileup or GATK)
   3. Run snpEffect, compute the "indexAlt" column and extract that index to it's own file
      • I use the workflow a workflow to run SOP: index snpEffect with Sample name, which actually computes several other files and indices need for building an SNP vs Samples grid.

chrLAB	0	.	chrLAB	JH03_B8M2	0	.	JH03_B8M2	JH03_B8M2	JH03_B8M2	JH03_B8M2	JH03_B8M2	chrLAB:0:chrLAB	chrLAB:0:chrLAB:JH03_B8M2
chrI	2323	.	C	T	471.72	.	AC=1;AF=0.50;AN=2;BaseQRankSum=0.330;DP=234;Dels=0.00;FS=21.822;HRun=2;HaplotypeScore=4.4329;MQ=44.55;MQ0=0;MQRankSum=-10.441;QD=2.02;ReadPosRankSum=0.083;EFF=DOWNSTREAM(LOW|||YAL067C|CALC_BIOTYPE||YAL067C|),DOWNSTREAM(LOW|||YAL068
W-A|CALC_BIOTYPE||YAL068W-A|),DOWNSTREAM(LOW|||YAL069W|CALC_BIOTYPE||YAL069W|),UPSTREAM(LOW|||YAL067W-A|CALC_BIOTYPE||YAL067W-A|),UPSTREAM(LOW|||YAL068C|CALC_BIOTYPE||YAL068C|)	GT:AD:DP:GQ:PL	0/1:175,58:234:99:502,0,6142	DOWNSTREAM(LOW|||YAL067C|CALC_BIOTY
PE||YAL067C|),DOWNSTREAM(LOW|||YAL068W-A|CALC_BIOTYPE||YAL068W-A|),DOWNSTREAM(LOW|||YAL069W|CALC_BIOTYPE||YAL069W|),UPSTREAM(LOW|||YAL067W-A|CALC_BIOTYPE||YAL067W-A|),UPSTREAM(LOW|||YAL068C|CALC_BIOTYPE||YAL068C|)	0/1	chrI:2323:C	chrI:2323:C:T

1. Concanate idxAlt files from all samples into one file

chrLAB:0:chrLAB:JH01_B8M1
chrI:2323:C:T
chrI:2331:A:C
chrI:3981:A:T
...

1. Group on c1, computing count(c1)
   • this produces one line for every SNP in any sample, with a count of how many samples it appears in
2. Filter to select only records where count()=num_samples

2micron:265:G:A	10
chrI:100399:G:C	10
chrI:101282:C:A	10

1. For each sample, remove the common SNP rows
   • I use the workflow SOP: VCF_fullidx remove common SNPs
   1. Text > Compare two data sets
      • Compare: snpEff+fullIdx [sample] (list of all variants for sample)
      • Using: c14 (the idxAlt)
      • Against: idxAlt with count=num_samples (common variants)
      • and column: c1 (the idxAlt)
   2. Trim off the index rows to get back to a VCF
      • Text > cut > "c1,c2,c3,c4,c5,c6,c7,c8,c9,c10,c11"
      • or any other subset of fields you want to report on