fix: update parameters and callsets #34
Conversation
FelixMoelder
changed the title
| @@ -11,14 +11,20 @@ __definitions__: | |||
| samples = params.samples.set_index("alias") | |||
| if "ffpe" not in samples.columns: | |||
| samples["ffpe"] = pd.NA | |||
| - sex = samples.loc["tumor", "sex"] | |||
| - sex = samples.loc[["tumor"], "sex"] | |||
There was a problem hiding this comment.
In case of groups with just one entry sample.loc["tumor", "sex"] will just return sex as a string.
But if there are multiple entries for a group sex will become a series.
In the previous implementation rendering the scenario only worked for groups with a single entry.
Changing sex to sample.loc[["tumor"], "sex"] will always return a series allowing to render single and multiple entries correctly.
Edit: In your other comment you mentioned that each alias should only occur once. So if we handle multiple panels by prefix this change probably also becomes unnecessary.
| if len(samples.loc[["tumor"], "ffpe"].unique()) != 1: | ||
| raise ValueError(f"All samples within a group must to be either ffpe or not.") | ||
| - | | ||
| if len(samples.loc[["tumor"], "purity"].unique()) != 1: |
There was a problem hiding this comment.
Each alias should occur only once in a group. We should also check for that when validating the sample sheet. If there are two panels for a patient we could name the two tumors tumor_panelname1 and 2. the scenario could support that by looking for the prefix tumor.
As we use panel data having high coverage for the mtb the preconfigured max read depth for varlociraptor preprocess was to low. To get a correct estimation the max-depth has been set to
30000in theconfig.yaml.Also considering the read position bias lead to missing variants in the past and therefore will be omitted.