Romeo
v2.0.0
Please log in
User Name
Password
Log in
Invalid username or password!
Log out
Data
GeneMapper files (.txt)
Browse...
?
The GeneMapper (fragment-analysis) peak-table export(s): one row per called peak, with its size (bp), height and area. A single file can hold many samples; you can also select several files at once (e.g. one per plate or per sample) and they are combined. Samples are matched to the settings by their sample/file name, so those must agree.
Settings (.xlsx)
Browse...
?
An Excel workbook describing each sample, one row per sample: the channel to read, analysis range(s), sizing (flank, repeat unit, correction), floor, group, time, sample_id (shared by technical replicates), controls and shared samples. Only one settings file is used; its sample names must match the GeneMapper files. See the Guide for the full column list.
Settings sheet
?
The name of the worksheet/tab to read inside the settings workbook (default 'Sheet1'). Change it if your settings are on a differently named tab.
Global overrides (optional)
?
Sizing: rpt = ((bp − flank) / repeat unit) + correction. Set any of these to apply that value to every sample, overriding its settings column; leave blank to use the settings value.
Flank
?
Flank – the constant subtracted from a peak's fragment size (bp) before dividing by the repeat unit, accounting for the fixed flanking sequence around the repeat. Overrides the settings flank for every sample when set; blank uses each sample's settings value.
Repeat unit
?
Repeat unit – the length in bp of one repeat (e.g. 3 for a CAG/CTG triplet). The flank-adjusted size is divided by this to convert base pairs to a repeat count. Overrides the settings value for every sample when set.
Correction
?
Correction – a constant added to the computed repeat number to calibrate sizing against a known-length standard (can be negative). Overrides the settings value for every sample when set.
Floor
?
The floor decides which data points are kept when computing a sample's distribution and instability – points whose signal is below it are treated as noise and dropped. 'Signal' means a peak's height or area (the analysis computes both). Set a value here to override the settings floor for every sample; leave blank to use each sample's settings floor. If a sample has no settings floor and nothing is set here, no floor is applied (all data points are kept). Below 1 = a fraction of the modal peak (e.g. 0.1 = 10% of the modal peak); 1 or more = an absolute signal value, applied separately to height and area – and area is typically ~6x the height, so the same absolute number bites much less on area.
Peak detection & instability
Peak detection window (odd number)
?
How many adjacent data points the local-maximum search spans to call a peak. A data point is one capillary scan – the sequencer records many scans per second, so a single peak is made of several data points (not the same as repeat units). Must be an odd number so the window has a centre; an even value is rounded up. Larger = smoother, fewer peaks.
Setpoint band width (repeats)
?
The central band, setpoint +/- half this width (in repeats), counted as 'at the setpoint'. Peaks above it are expansions, below it contractions; sets the expansion / contraction / tail ratios. Both setpoints (control repeat length and sample's own mode) are computed; pick which to view on the Instability and Trends tabs.
Quality control
Minimum modal peak height (0 = off)
Minimum modal peak area (0 = off)
?
Judged per sample, per analysis range. For a given sample and range: if its modal peak height is below the height minimum, or its modal peak area is below the area minimum, that sample-range is flagged 'below_minimum' (height and area are judged separately) and kept out of its group's control. 'Signal' here is the modal peak's height or area. 0 = off. Detail in the Guide.
Drop flagged sample-ranges before analysis
?
The only step that removes data. When ticked, any single sample-range measurement flagged 'below_minimum' (its modal peak is below the minimum you set) or 'no_peak' (no usable peak found) is dropped. It is judged per signal and per range – a sample's weak height in one range can be dropped while its area in that range is kept, and it never drops a range for every sample. Dropped measurements leave technical-replicate collapse and everything downstream (trends); the distribution and instability tables still keep every row, each carrying a qc_flag column. With both minimums off, only 'no_peak' ranges are dropped.
Technical replicate outliers
?
Technical replicates are files sharing a sample_id within an analysis range – repeat measurements of one biological sample. This section detects replicates whose sizing disagrees with their siblings; collapsing them is set in the next section.
Detect technical-replicate outliers
?
Master switch for outlier handling (off by default). When on, the options below appear, replicates are tested for outliers, and the technical_replicates audit is produced (a sheet and a QC-tab table). When off, none of that runs. Either way, replicates are still collapsed for the by-sample views using the method below.
Outlier detection metric
Modal repeat length
Median repeat length
Mean repeat length
?
Which per-sample metric flags an outlier replicate. A replicate is flagged when this metric sits unusually far from the others sharing its sample_id – more than k interquartile ranges beyond their quartiles (k is set below). Tested on height and area separately; needs at least 3 replicates.
Outlier sensitivity (k)
?
How far a value must sit from the other replicates to count as an outlier: k × their interquartile range beyond the quartiles. Lower k flags more; 1.5 is the usual default. Needs at least 3 replicates.
Remove technical outliers
?
When on, replicates flagged as outliers are excluded wherever replicates are combined into one value per sample: the collapsed by-sample_id tables and stats, and the group control. Per-file tables keep every replicate, with outliers flagged (not removed). Off by default – flag only.
Technical replicate collapse
?
Collapsing reduces a sample's technical replicates to one value per biological sample. Separate from outlier detection above.
Collapse technical replicates
?
When off, nothing is collapsed – every file is kept as its own sample everywhere: the group control uses every file, and the Distribution, Instability and Trends tabs show per file only. When on, replicates are collapsed to one value per sample_id wherever that applies: the group control, the collapsed-by-sample versions of the Distribution and Instability tables, and the Trends stats. Off by default.
Collapse method
mean
median
mode
?
How technical replicates are combined into one value per biological sample: mean, median or mode across the reps (after any outlier removal).
Shared samples
Share samples across groups (from settings)
?
Share flagged samples across groups – e.g. one control or t = 0 measured once and used by several arms. shared_sample flags the rows to share; shared_family names the set of groups that share them. Those rows are copied into each group in the family that doesn't already have its own, before controls and trends. Group-scoped: blank family = not shared; a different family = a separate set. Detail in the Guide.
Controls
Control source
From settings (manual values)
Calculate by group
?
Where the 'control' index setpoint comes from. 'From settings' uses the manual_control_repeat_length column; 'Calculate by group' derives it from each group's samples (options below). Grouping uses the settings group column; if no group is set, all samples are treated as one group. If a sample/range has no control value its 'control' metrics are blank; 'mode' is always available.
Identify control samples from
earliest timepoint
flagged sample(s)
all samples in group
?
Which of each group's samples form its control set: the earliest timepoint, the sample(s) marked group_control_sample, or all samples in the group.
Signal
height
area
?
Which signal the control repeat length is measured on.
Control metric
Modal repeat length
Mean repeat length
Median repeat length
?
Which repeat-length metric of each control sample is used: the modal peak, the mean or the median repeat length.
Combine across controls by
mean
median
mode
?
How the per-sample control values are combined into one reference repeat length per group: mean, median or mode. Reads as: from the identified samples, take the chosen metric on the chosen signal, combined by this. Whether the control first collapses each sample's technical replicates to one value is set by 'Collapse technical replicates' above (off = every file contributes).
Run analysis
Running analysis…
Running analysis – you can keep using the app.