SequenceCoordinateConverter (SCC)

A tool for converting genomic coordinates between different versions of the same species or between closely related species.

Method

--------|---------reference 1----------------- chr1A
    ----|-read1----->
    <---|--read2-----
      --|---read3----->

## "|" represents the pos need to be converted

-----------------reference 2------/----------- chr1A
                              ----/-read1----->
                              <---/--read2-----
                                --/---read3----->

## "/" represents the pos has been converted

When we have a position (Pos1) in reference 1, we can find the reads that cover this position, then we can get the alignment position in reference 2 of these reads, and we can calculate the Pos2 in reference 2 that corresponds to the Pos1 in reference 1.

Usage

1. Coordinate Conversion for the Same Species with Different Assembly Versions

a. Align the second-generation reads of the species to both reference1 and reference2 using bwa mem.

bwa mem reference1.fa reads.fq > alignment1.sam
bwa mem reference2.fa reads.fq > alignment2.sam

b. Sort the alignment file for reference1 based on genomic coordinates and create an index using samtools.

samtools sort -o alignment1_sorted.bam -O bam alignment1.sam
samtools index alignment1_sorted.bam

c. Sort the alignment file for reference2 based on the read name.

samtools sort -n -o alignment2_namesort.bam -O bam alignment2.sam

d. Use bri (https://github.com/jts/bri) to index the sorted alignment2.

bri index alignment2_namesort.bam

e. Use SCC for position conversion.

$ SCC convert -h
usage: SCC convert [-h] -b BAM1 -B BAM2 -r REF1 -R REF2 -s SNP [-t THREAD] -o OUTPREFIX [--use-server]

SequenceCoordinateConverter (SCC) / fengcong@caas.cn

optional arguments:
  -h, --help            show this help message and exit
  -b BAM1, --bam1 BAM1  Bam/cram file of reference 1
  -B BAM2, --bam2 BAM2  Bam file of reference 2
  -r REF1, --ref1 REF1  Fasta file of reference 1
  -R REF2, --ref2 REF2  Fasta file of reference 2
  -s SNP, --snp SNP     SNP position file of reference 1, format: chr1A:100,
                        one line one SNP
  -t THREAD, --thread THREAD
                        Thread number
  -o OUTPREFIX, --outprefix OUTPREFIX
                        Output prefix
  --use-server          Try to use server mode if available

2. Server Mode for High-throughput Processing

When processing thousands of tasks that use the same reference files, loading the index for each task wastes memory and time. SCC now includes a server mode that loads the index once and shares it through shared memory.

a. Start the server:

SCC server -B alignment2_namesort.bam

You can also run it as a daemon in the background:

SCC server -B alignment2_namesort.bam -d

b. Run conversion tasks with the --use-server flag:

SCC convert -b alignment1_sorted.bam -B alignment2_namesort.bam -r reference1.fa -R reference2.fa -s positions.txt -o output --use-server

c. When finished, stop the server:

SCC stop-server

3. Coordinate Conversion Between Different Species or Closely Related Species

a. Choose the reads of either reference1 or reference2 and align them to both references using bwa mem.

bwa mem reference1.fa chosen_reads.fq > alignment1.sam
bwa mem reference2.fa chosen_reads.fq > alignment2.sam

... (Rest of the steps would be similar to the first scenario)

Results

Example result:

Ref1_chr	Ref1_pos	Ref1_base	Ref2_chr	Ref2_pos	Ref2_base	code
chr1B_part1	882559	A	Chr1B_part1	875080	A	0
chr1B_part1	58917715	G	Chr1B_part1	64796436	C	1

Explanation of the code Field

The code field is used to describe the relationship between Ref1 and Ref2 at the corresponding positions:

• 0: No change in assembly between Ref1 and Ref2 at this position.
• 1: Change in assembly orientation between Ref1 and Ref2 at this position.
• 2: Difference in assembly between Ref1 and Ref2 at this position, but the orientation remains unchanged.
• 3: Difference in assembly between Ref1 and Ref2 at this position, and the orientation has also changed.

Discussion

For the second scenario, it's not entirely clear whether one should use reads from reference1 or reference2 for the conversion, or perhaps both. This is an area that requires further investigation.

A side-by-side comparison or hybrid approach employing both sets of reads could offer more insights into this issue.