Documentation updates for 0.3.12 release (Illumina#110)

* Update RELEASES.md (Illumina#87)

* Updated RELASES.md and normalization.md (Illumina#109)

RELEASES.md

@@ -1,10 +1,26 @@
 # Hap.py Release Notes / Change Log
 
+## v0.3.12
+
+| Ticket  | Description                                                                        |
+|---------|------------------------------------------------------------------------------------|
+| HAP-355 | Ensure compatibility with genome VCF from DRAGEN pipelines                         |
+| HAP-356 | quantify module throws regex error                                                 |
+| HAP-357 | Deal with change in Java license terms                                             |
+| HAP-359 | Hap.py crashes when ingesting variants genotyped as <NON_REF>                      |
+| HAP-360 | Apply non-ref filter to truth set when --preprocess-truth is set                   |
+| HAP-361 | Hap.py "blocksplit" processes still failing on DRAGEN gVCFs even with pre-filtering NON_REF genotypes 
+| HAP-362 | Hap.py blocksplit sometimes crashes with "invalid next size (fast)" error          |
+| HAP-363 | Use 'git describe' to obtain version number during Cmake                           |
+
 ## v0.3.11
 
 | Ticket  | Description                                                                        |
 |---------|------------------------------------------------------------------------------------|
 | HAP-341 | Update rtgtools to 3.10.1                                                          |
+| HAP-342 | som.py output tables to quantify / GA4GH format                                    |
+| HAP-346 | src/sh/illumina-setup.sh out of date                                               |
+| HAP-352 | string formatting in som.py: last interval is always [1.0, 1.0] in *extended.csv   |
 
 ## v0.3.10
 

doc/normalisation.md

@@ -16,7 +16,8 @@ usage: VCF preprocessor [-h] [--location LOCATIONS] [--pass-only]
                         [-T TARGETS_BEDFILE] [-L] [-D] [--bcftools-norm]
                         [--fixchr] [--no-fixchr] [--bcf] [-v] [-r REF]
                         [-w WINDOW] [--threads THREADS] [--force-interactive]
-                        [--logfile LOGFILE] [--verbose | --quiet]
+                        [--logfile LOGFILE] [--verbose | --quiet] [--filter-nonref]
+                        [--convert-gvcf-truth] [--convert-gvcf-query]
                         input output
 ```
 
@@ -129,6 +130,30 @@ The default value is 10000, which should be sufficient for short reads.
                         that size are not expected to interfere).
 ```
 
+The presence of the <NON_REF> symbolic allele in genome VCFs can cause problems
+for hap.py, especially if it is part of a genotype. As a workaround, we 
+provide several options. Since variants genotyped as <NON_REF> cannot be
+sensibly scored, the we provide the following option, which is safe to use
+on both genome VCFs and standard VCFs:
+
+```
+  --filter-nonref       Remove any variants genotyped as <NON_REF>.
+                        
+```
+
+If hap.py still crashes when processing a genome VCF, we provide separate
+options to perform on-the-fly conversion of a genome VCF to a standard VCF
+by removing all <NON_REF> alleles and non-variant blocks. Note that this 
+also removes some fields from the INFO column. These options should only
+be used on genome VCFs since attempting to convert a standard VCF will 
+cause all biallelic variants to be filtered out (most of them).
+
+```
+  --convert-gvcf-truth Convert the truth genome VCF to a standard VCF.
+  --convert-gvcf-query Convert the query genome VCF to a standard VCF.
+                        
+```
+
 Runtime behaviour can also be controlled as follows:
 
 ```