ViralConsensus v0.0.6

• Added (hopefully) correct support for primers on the reverse strand (lines where column 2 > column 3 in the BED file)
  • Thanks, @RanWei1987!

ViralConsensus v0.0.5

• Updated "Is this larger than the minimum frequency?" comparisons for stability
  • I used to do if(count / tot_depth > min_freq)
  • I changed it to if(count > tot_depth * min_freq)
  • count and tot_depth used to be double due to the division, but I changed them to unsigned integers
    • min_freq was a double, but now it's a long double
• Changed all double to long double just in case floating point precision is an issue in edge cases
  • This is only relevant to comparisons against the minimum frequency to call bases/insertions
• Changed most int/long integer types to int??_t and uint??_t types
  • This probably won't impact anything, but figured I would do it just in case for cross-platform stability

ViralConsensus v0.0.4

• Open the reference genome of an input CRAM file properly
  • Originally, CRAM files were always pulling a remote copy of the reference genome (#6)
  • To fix it, I used the htslib cram_load_reference() function, which solved the immediate problem (0b99fca)
  • However, it seems like this is an internal htslib function, and the proper fix is to use the hts_set_opt() function instead
    • Thanks for the fix, @daviesrob! And thanks for your input, @jmarshall and @jkbonfield!
  • This shouldn't change anything from the user's perspective: it just makes it easier to compile ViralConsensus

ViralConsensus v0.0.3

• Fixed bug where local reference genome file is ignored when using CRAM file
  • Prior to v0.0.3, when the input file is a CRAM file, htslib tries to automatically download the reference genome based on the MD5 hash: it ignores the user-provided reference genome FASTA file
  • As a result, ViralConsensus would sometimes crash in specific contexts when the running environment was unable to connect to the internet
  • Now, this issue should be fixed: it should be using the local file and should be working in offline contexts

ViralConsensus v0.0.2

• Fixed bug when handling minority insertions with sufficiently high coverage (thanks @alexcritschristoph!)
  • Normally, if an erroneous insertion is encountered, the number of reads supporting that insertion should be low
  • However, if sequencing is deep enough, the number of reads supporting that insertion could exceed the "minimum coverage" threshold, even though there could be a ton of reads that don't have that insertion
    • E.g. if there are 100 reads with ACGT and 1,000,000 reads with ACT (no G), the ACGT would still appear in the consensus
  • To patch this, we now correctly count "no insertion" between matches/mismatches

ViralConsensus v0.0.1

• Initial release