Add support for SSM files within ICGC projects

[victorlin]

I'm wondering if it would be appropriate to add functionality of reading files such as simple_somatic_mutation.open.BRCA-US.tsv.gz.

Reading this data is the initial step of a project I will be starting soon. I would be more than willing to implement the parser.

For reference, the file has these columns:

1. icgc_mutation_id
2. icgc_donor_id
3. project_code
4. icgc_specimen_id
5. icgc_sample_id
6. matched_icgc_sample_id
7. submitted_sample_id
8. submitted_matched_sample_id
9. chromosome
10. chromosome_start
11. chromosome_end
12. chromosome_strand
13. assembly_version
14. mutation_type
15. reference_genome_allele
16. mutated_from_allele
17. mutated_to_allele
18. quality_score
19. probability
20. total_read_count
21. mutant_allele_read_count
22. verification_status
23. verification_platform
24. biological_validation_status
25. biological_validation_platform
26. consequence_type
27. aa_mutation
28. cds_mutation
29. gene_affected
30. transcript_affected
31. gene_build_version
32. platform
33. experimental_protocol
34. sequencing_strategy
35. base_calling_algorithm
36. alignment_algorithm
37. variation_calling_algorithm
38. other_analysis_algorithm
39. seq_coverage
40. raw_data_repository
41. raw_data_accession
42. initial_data_release_date

[Ad115]

Sorry for the late response. Of course! It would be great. Feel free to make the changes and send me a pull request! 💃

[Ad115]

Also, it would be great if you used the facilities in the standard library for gzip files and tsv files.

Maybe that parser for project SSM's may be another class analogous to the SSM_Reader, but how would it be named?? Maybe Project_SSM_Reader? or should we extend the interface of the existing parser?? I'm thinking something like:

reader = SSM_Reader(filename='simple_somatic_mutation.open.BRCA-US.tsv.gz', file_type='project ssm')

Another thing that seems sensible to me would be to refactor the dependency on the vcf.Reader, so that one could change from vcf to tsv and not make a separate class. Something like:

# The old behavior:
reader = SSM_Reader(filename=' simple_somatic_mutation.aggregated.vcf.gz', file_type='vcf')

# The new behavior:
reader = SSM_Reader(filename='simple_somatic_mutation.open.BRCA-US.tsv.gz', file_type='tsv')

And in each case, internally, a different reader (vcf.Reader or csv.reader) would be instantiated internally.

I'd be great to hear your thoughts on the subject 🌝

[victorlin]

It seems like ICGC provides SSM data in two formats: VCF-like and ICGC-like Mutation Format. I'll reference the "ICGC-like" format as TSV for now.

It's worth noting that the TSV format isn't only available per-project. All SSM data downloaded from the web portal is in the TSV format. It seems to be a more widely available option, whereas the VCF format is only available by downloading all at once from the data release in DCC/current/Summary.

That is just my personal understanding of the ICGC structure. Maybe this library could automatically detect either format:

reader1 = SSM_Reader(filename='simple_somatic_mutation.aggregated.vcf.gz')
reader2 = SSM_Reader(filename='simple_somatic_mutation.open.BRCA-US.tsv.gz')

or have the user specify which is passed in:

reader1 = SSM_Reader(vcf='simple_somatic_mutation.aggregated.vcf.gz')
reader2 = SSM_Reader(tsv='simple_somatic_mutation.open.BRCA-US.tsv.gz')

[Ad115]

I love the idea of automatic detection of the file format, given that every well-formed VCF must start with a line specifying the format, according to the VCF specification.

Given that, I can not think of a plausible case for having the user to specify the file format manually, maybe it would be good to have an optional switch, or to remove the option altogether.