This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Tool: DiscoSnp++ 2.1.2 release: now genotypes and creates VCFs

We are pleased to announce a new release of the discoSnp++ tool (Home page: http://colibread.inria.fr/software/discosnp/)

DiscoSNP++ is a reference-free SNP/indel discovery tool.

From version 2.1.2:

  1. discoSnp++ generates a VCF as output:
    • Without mapping positions if no reference genome is available
    • With mapping positions else. In this latter case, discoSnp++ uses bwa for mapping.
  2. discoSnp++ computes genotypes from predicted coverages of variants. This predictions are reported both in the fasta output and in the VCF file.

As usual any comment or feedback (negative or positive :)) is welcome.

Pierre

snp discosnp genotyping indel
• 3,987 views • link
updated by Ram Ram • written by pierre.peterlongo pierre.peterlongo

Hello, compilation of latest version fails for some reason (first time I try to compile). Any ideas? http://pastebin.ca/2962771 I am running centos 64bit

• 0 views • link
updated by Ram Ram • written by Adrian Pelin Adrian Pelin

Hi,

Are you using clang ?

In this case, could you test to use gcc instead?
This would need to avoid to use the automatic compiler (./compile_discoSnp++.sh). For instance if you disposes from gcc 4.9 installed;

rm -rf build
mkdir build
cd build
cmake -DCMAKE_C_COMPILER=gcc-4.9 -DCMAKE_CXX_COMPILER=g++-4.9 ..
make
cd ..

Best, Pierre

• 0 views • link
written by pierre.peterlongo pierre.peterlongo

Hello, that worked! after cmake

cmake -DCMAKE_C_COMPILER=/opt/centos/devtoolset-1.0/root/usr/bin/gcc -DCMAKE_CXX_COMPILER=/opt/centos/devtoolset-1.0/root/usr/bin/g++ ..

i did make. Where can I find binaries? Thanks!

• 0 views • link
written by Adrian Pelin Adrian Pelin

Nice to know, thanks. Binaries are in the ROOT/build/tools/ directory. However, note that the principal script is ROOT/run_discoSnp++.sh.

Pierre

• 0 views • link
updated by Ram Ram • written by pierre.peterlongo pierre.peterlongo

Quick question (is it ok to ask questions here or should I make a new post?). I am a bit confused about the difference between variation found in the coherent and uncoherent files. I thought it was really based on coverage. I have the following example of 2 sets of coherent sequences:

>SNP_higher_path_20|P_1:30_T/G|high|nb_pol_1|C1_114|C2_97|C3_96|C4_92|C5_88|G1_0/0:9,347,2284|G2_0/0:9,296,1944|G3_0/0:8,293,1924|G4_0/0:12,251,1808|G5_0/1:144,133,1581|Q1_71|Q2_70|Q3_70|Q4_68|Q5_66|rank_0.32756
>SNP_lower_path_20|P_1:30_T/G|high|nb_pol_1|C1_0|C2_0|C3_0|C4_2|C5_16|G1_0/0:9,347,2284|G2_0/0:9,296,1944|G3_0/0:8,293,1924|G4_0/0:12,251,1808|G5_0/1:144,133,1581|Q1_0|Q2_0|Q3_0|Q4_47|Q5_61|rank_0.32756

>SNP_higher_path_178|P_1:30_A/C|high|nb_pol_1|C1_7|C2_10|C3_21|C4_3|C5_0|G1_1/1:1644,187,48|G2_1/1:1653,170,76|G3_0/1:1907,152,191|G4_1/1:2071,279,16|G5_1/1:2044,311,9|Q1_58|Q2_57|Q3_51|Q4_49|Q5_0|rank_0.22405
>SNP_lower_path_178|P_1:30_A/C|high|nb_pol_1|C1_87|C2_89|C3_107|C4_106|C5_102|G1_1/1:1644,187,48|G2_1/1:1653,170,76|G3_0/1:1907,152,191|G4_1/1:2071,279,16|G5_1/1:2044,311,9|Q1_65|Q2_61|Q3_63|Q4_67|Q5_68|rank_0.22405

and then these 2 sets are considered uncoherent:

>SNP_higher_path_69|P_1:30_T/G|high|nb_pol_1|C1_13|C2_14|C3_9|C4_3|C5_0|G1_1/1:1393,123,115|G2_1/1:1520,134,123|G3_1/1:1761,190,64|G4_1/1:1614,213,18|G5_1/1:2164,329,9|Q1_58|Q2_62|Q3_57|Q4_59|Q5_0|rank_0.21337
>SNP_lower_path_69|P_1:30_T/G|high|nb_pol_1|C1_77|C2_84|C3_94|C4_83|C5_108|G1_1/1:1393,123,115|G2_1/1:1520,134,123|G3_1/1:1761,190,64|G4_1/1:1614,213,18|G5_1/1:2164,329,9|Q1_65|Q2_64|Q3_66|Q4_68|Q5_68|rank_0.21337

>SNP_higher_path_39|P_1:30_T/G|high|nb_pol_1|C1_12|C2_13|C3_13|C4_2|C5_0|G1_1/1:1615,155,98|G2_1/1:1761,169,105|G3_1/1:1645,154,108|G4_1/1:1748,242,12|G5_1/1:1724,263,8|Q1_51|Q2_61|Q3_60|Q4_43|Q5_0|rank_0.19546
>SNP_lower_path_39|P_1:30_T/G|high|nb_pol_1|C1_88|C2_96|C3_90|C4_89|C5_86|G1_1/1:1615,155,98|G2_1/1:1761,169,105|G3_1/1:1645,154,108|G4_1/1:1748,242,12|G5_1/1:1724,263,8|Q1_66|Q2_66|Q3_65|Q4_67|Q5_68|rank_0.19546

Do these provide any clue as to why some are coherent and why some are not? Because when I look at the coverage of the minimal variant it is not very obvious to me.

Thank you!

• 0 views • link
updated by Ram Ram • written by Adrian Pelin Adrian Pelin

(A new post would be best for this question.. but it's alright!)

I'll let Pierre confirm the following guess: coherent/incoherent is indeed based on coverage, but one should distinguist k-mer coverage with k-read-coverage (as defined is discosnp paper). A SNP can be k-read-incoherent, meaning it might be an assembly artefact that the reads cannot explain, yet the whole path might have a sufficient mean coverage of individual kmers.

• 0 views • link
written by Rayan Chikhi Rayan Chikhi

1 answer

Hi Adrian

The read coherency is computed as follows:

The reads of each read set are mapped back on each prediction (authorizing by default one mismatch anywhere but not on the variant(s) position(s).

For each read set, each predicted sequence is said "k-read-covered" if all kmers of this sequence are covered by at least c reads (c being a main parameter, =4 by default).

Variants for which the two sequences are not k-read-coherent for all read sets are declared "uncoherent".

The read coverage indicated in the outputs (.fa and .vcf) are the sum of number of the read mapped. Thus this is possible to have a high read coverage for uncoherent variants: this means that a lot of reads mapped some parts of the sequence but that some other regions of the sequence are not mapped.

I hope this answers your questions.

Best, Pierre.

• 1 views • link
updated by Ram Ram • written by pierre.peterlongo pierre.peterlongo

Log in to answer this question.

More posts like this