unresolved sites after using bcftools +fixref

[hyl317]

Hi, I need to fix the REF allele for my .vcf file per Sanger Imputataion Server's requirement. My .vcf happens to contain dbSNP reference identificators, so I followed the command here http://samtools.github.io/bcftools/howtos/plugin.fixref.html. However, the bcftools gives me the following message:

[yh362@cbsubscb10 MS_ALL]$ bcftools +fixref MS_ALL_illumina_QC2.vcf -Ov -o fixref.vcf -- -d -f /bscb/data/human_reference/GRCh37/human_g1k_v37/human_g1k_v37.fasta -i dbsnp_138.b37.vcf.gz

Warning: corrected position(s) results in unsorted VCF, for example 2:95398090 comes after 2:96007302
         The standard unix `sort` or `vcf-sort` from vcftools can be used to fix the order.

# SC, guessed strand convention
SC      TOP-compatible  0
SC      BOT-compatible  0

# ST, substitution types
ST      A>C     38564   8.4%
ST      A>G     172187  37.6%
ST      A>T     0       0.0%
ST      C>A     43618   9.5%
ST      C>G     0       0.0%
ST      C>T     0       0.0%
ST      G>A     203611  44.5%
ST      G>C     0       0.0%
ST      G>T     0       0.0%
ST      T>A     0       0.0%
ST      T>C     0       0.0%
ST      T>G     0       0.0%

# NS, Number of sites:
NS      total           457980
NS      ref match       158794  34.7%
NS      ref mismatch    299186  65.3%
NS      flipped         0       0.0%
NS      swapped         69948   15.3%
NS      flip+swap       0       0.0%
NS      unresolved      229317  50.1%
NS      fixed pos       5       0.0%
NS      skipped         0
NS      non-ACGT        0
NS      non-SNP         0
NS      non-biallelic   0

So it appears that half of the sites remain unresolved. What might be the reason for this? Thanks!

[pd3]

It is probably because there are many missing IDs or non-matching IDs in the VCF.

[ruxi]

@hyl317: are you using affy genomewide5 array? I'm getting similar numbers