Filterscan Report

Filterscan Report: Fields

Name	Data Type
Sample	str
Branches	dict[str, str]
Reference	str
Region	str
Make each tile this length (if 0, use 2x the median read length)	int
Make adjacent tiles overlap by at least this fraction of length	float
Erase the filter reports/batches from the tiling step	bool
Consider only pairs of positions no more than this far apart when finding domains (if 0, do not restrict pairs beyond what the tile length already bounds)	int
Analyze only pairs of positions whose number of jointly covering reads is at least this value when finding domains (pairs with less coverage are too noisy to score reliably)	int
Analyze only pairs of positions whose expected number of reads mutated at both positions (under the assumption that the positions mutate independently) is at least this value when finding domains (standard practice for a chi-square test, which becomes unreliable when an expected count drops below about 5)	float
Count a pair of positions as a domain-boundary-crossing bridge only if its anti-correlation clears this false discovery rate (FDR), Benjamini-Hochberg-adjusted over every analyzed pair’s exact hypergeometric p-value: higher values call more (and weaker) bridges; lower values call fewer, more conservative ones (this value also seeds the initial background bridge rate before it is re-estimated from the pairs outside the first-pass domains)	float
Count a pair of positions as a domain-boundary-crossing bridge only if its expected-to-observed both-mutated fold change is at least this value (reflects the underlying cluster split, e.g. a 50/50 split depletes joint mutations differently than a 90/10 split): higher values require a larger effect to call a bridge	float
Call a region a domain only if its enrichment in bridge pairs clears this false discovery rate (FDR), Benjamini-Hochberg-adjusted over every candidate region’s exact binomial p-value against the background bridge rate: higher values call more (and weaker) domains; lower values call fewer, more conservative ones	float
Merge two adjacent domains only if the bridge pairs crossing every position between them are enriched at this false discovery rate (FDR), Benjamini-Hochberg-adjusted over every candidate crossing point’s exact binomial p-value against the background bridge rate: higher values merge more (joining domains connected by weaker long-range pairs); lower values merge fewer	float
Grow each called domain into the gaps on either side of it, up to max-domain-length in total length (runs before –fill): if the gap between two domains is smaller than their combined remaining budget, it closes entirely; otherwise each domain grows as much as its own budget allows and any leftover space stays a gap (which –fill can then fill). Domains that grew are recorded as ‘widened’ and unchanged ones as ‘original’ in the report	bool
Insert additional domains into every gap – leading, interior, and trailing (after –widen has run, if set) – so every position in the scanned region ends up in exactly one domain: a gap no longer than max-domain-length becomes a single domain; a longer gap is split into the minimum number of domains of as equal length as possible, none exceeding max-domain-length. Inserted domains are recorded as ‘filled’ in the report	bool
Bound every candidate domain to at most this many positions when finding domains, so no domain exceeds the length that clustering can use (if 0, use 2x the median read length)	int
Keep only domains with at least this many positions (drops the shortest called blocks; skipped when –widen is set, since widening grows short domains into neighboring space instead of dropping them)	int
Coordinates of tiles (end5, end3)	list[list[int]]
Number of bridge pairs (anti-correlated, significant, large-effect)	int
Estimated background (null) rate of bridge pairs among eligible pairs	float
Number of domains detected	int
Domains found, mapped from (end5, end3) to how each was produced: ‘original’ (called/determined as detected), ‘widened’ (–widen grew it into a gap), or ‘filled’ (–fill inserted it into a gap)	dict[str, str]
Time began	str
Time ended	str
Time taken (minutes)	float
Version of SEISMIC-RNA	str

Filterscan Report: Example

{
    "Sample": "sample1",
    "Branches": {
        "idmut": "",
        "filterscan": ""
    },
    "Reference": "myref",
    "Region": "full",
    "Make each tile this length (if 0, use 2x the median read length)": 360,
    "Make adjacent tiles overlap by at least this fraction of length": 0.5,
    "Erase the filter reports/batches from the tiling step": true,
    "Consider only pairs of positions no more than this far apart when finding domains (if 0, do not restrict pairs beyond what the tile length already bounds)": 0,
    "Analyze only pairs of positions whose number of jointly covering reads is at least this value when finding domains (pairs with less coverage are too noisy to score reliably)": 1000,
    "Analyze only pairs of positions whose expected number of reads mutated at both positions (under the assumption that the positions mutate independently) is at least this value when finding domains (standard practice for a chi-square test, which becomes unreliable when an expected count drops below about 5)": 5.0,
    "Count a pair of positions as a domain-boundary-crossing bridge only if its anti-correlation clears this false discovery rate (FDR), Benjamini-Hochberg-adjusted over every analyzed pair's exact hypergeometric p-value: higher values call more (and weaker) bridges; lower values call fewer, more conservative ones (this value also seeds the initial background bridge rate before it is re-estimated from the pairs outside the first-pass domains)": 0.05,
    "Count a pair of positions as a domain-boundary-crossing bridge only if its expected-to-observed both-mutated fold change is at least this value (reflects the underlying cluster split, e.g. a 50/50 split depletes joint mutations differently than a 90/10 split): higher values require a larger effect to call a bridge": 2.0,
    "Call a region a domain only if its enrichment in bridge pairs clears this false discovery rate (FDR), Benjamini-Hochberg-adjusted over every candidate region's exact binomial p-value against the background bridge rate: higher values call more (and weaker) domains; lower values call fewer, more conservative ones": 0.05,
    "Merge two adjacent domains only if the bridge pairs crossing every position between them are enriched at this false discovery rate (FDR), Benjamini-Hochberg-adjusted over every candidate crossing point's exact binomial p-value against the background bridge rate: higher values merge more (joining domains connected by weaker long-range pairs); lower values merge fewer": 0.05,
    "Grow each called domain into the gaps on either side of it, up to max-domain-length in total length (runs before --fill): if the gap between two domains is smaller than their combined remaining budget, it closes entirely; otherwise each domain grows as much as its own budget allows and any leftover space stays a gap (which --fill can then fill). Domains that grew are recorded as 'widened' and unchanged ones as 'original' in the report": true,
    "Insert additional domains into every gap -- leading, interior, and trailing (after --widen has run, if set) -- so every position in the scanned region ends up in exactly one domain: a gap no longer than max-domain-length becomes a single domain; a longer gap is split into the minimum number of domains of as equal length as possible, none exceeding max-domain-length. Inserted domains are recorded as 'filled' in the report": true,
    "Bound every candidate domain to at most this many positions when finding domains, so no domain exceeds the length that clustering can use (if 0, use 2x the median read length)": 50,
    "Keep only domains with at least this many positions (drops the shortest called blocks; skipped when --widen is set, since widening grows short domains into neighboring space instead of dropping them)": 20,
    "Coordinates of tiles (end5, end3)": [
        [
            1,
            180
        ]
    ],
    "Number of bridge pairs (anti-correlated, significant, large-effect)": 62,
    "Estimated background (null) rate of bridge pairs among eligible pairs": 0.002909090909090909,
    "Number of domains detected": 6,
    "Domains found, mapped from (end5, end3) to how each was produced: 'original' (called/determined as detected), 'widened' (--widen grew it into a gap), or 'filled' (--fill inserted it into a gap)": {
        "1,3": "filled",
        "4,53": "widened",
        "54,88": "filled",
        "89,122": "filled",
        "123,172": "widened",
        "173,180": "filled"
    },
    "Time began": "2026-07-30 at 02:38:39",
    "Time ended": "2026-07-30 at 02:39:30",
    "Time taken (minutes)": 0.85,
    "Version of SEISMIC-RNA": "0.26.0dev"
}