Comparing two datasheets without copying them

The symptom is a pair of PDFs that will not line up. One antibody sheet leads with a dilution picture. The other leads with a clone name and a storage sentence. Both look complete, and a meeting spent scrolling them will not tell you which product the assay can accept. The fix is to stop reading the sheets as tables and to extract the same fields into a table the laboratory owns. That habit sits downstream of how to write a laboratory sourcing enquiry. Bring the finished table to laboratory procurement, or start from academic research if the buyer is a university group. A scope can be discussed through a quotation request or contact.

Why the sheets refuse to compare

A datasheet is a manufacturer's argument, arranged in the order that argument prefers. It is allowed to emphasise sensitivity, host species, or a pretty western blot. It is not obliged to use your column headings. When you copy the manufacturer's table into your note, you also copy the omissions and the order. Two copied tables then sit side by side as if they were the same instrument, which they are not.

Your table is a specification in miniature. Each row is a field you already decided matters. Each column is one candidate product. An empty cell is a question, not a failure of formatting. NIST laboratory metrology is a public reminder that a measurement needs defined conditions. Analyte, matrix, and range are those conditions for a kit or a detection reagent. Leaving them inside someone else's layout hides whether the conditions match your samples.

The six fields worth extracting

Use the same six fields for both sheets, even when one sheet buries a field in a footnote.

Analyte is what the reagent or kit claims to detect or to act on. Write the molecule, the epitope or the enzyme's substrate class, in your words. "Phospho-protein" is too wide if you care about one site. Name the site.

Matrix is the material the claim was built on: lysate, serum, purified protein, a particular cell line, formalin-fixed tissue. Your samples are a matrix too. If the sheet's matrix is not yours, the cell stays filled with their matrix and you add a note that yours is unshown.

Range is the concentrations, sizes, or activities over which the claim is stated. Record the unit exactly. Do not convert a qualitative "high" into a number.

What the control was is the comparison the manufacturer used: a positive lysate, a recombinant protein, an isotype, a no-template reaction, a blank matrix. This field is where sheets quietly diverge. One antibody was shown against a knockout lysate. Another was shown against a blocking peptide. Those are different controls. Write both, and do not call them the same word "validated".

Storage is the temperature class and any light or freeze-thaw limit the sheet states. Copy the range they printed into your cell, and do not upgrade "refrigerate" into a number they did not print.

Documents are what you can ask to receive with the lot: a certificate of analysis, a safety data sheet, a clone statement, a calibration record. A figure in the datasheet is not a document you will hold for the lot that arrives next month.

Trademarked product names may appear in your column headers so you know which PDF you read. The cells stay in your words. You are not reproducing the manufacturer's page.

A worked extraction for two antibody sheets

Suppose you need an antibody against a phosphorylated site on a signalling protein, for a western blot of a cultured-cell lysate. Sheet A opens with a picture and the words "predicted to react". Sheet B opens with a clone identifier and a dilution. Neither layout matches.

You rule a table with the six fields and two columns. Analyte: you write the protein and the residue for both, and you notice Sheet A never names the residue while Sheet B does. Matrix: Sheet A says paraffin sections; Sheet B says whole-cell lysate. Your samples are lysates, so Sheet A's claim is about a matrix you are not running. Range: Sheet A gives none; Sheet B gives a lysate load in micrograms of protein per lane. You record "not stated" for A and the load, with its unit, for B. Control: Sheet A shows a peptide-block experiment; Sheet B shows a treated versus untreated lysate. You write those phrases and you do not merge them into "specificity tested". Storage: both say a frozen class; only Sheet B mentions aliquotting and a limit on freeze-thaw cycles. Documents: Sheet A offers a general data page; Sheet B's sheet mentions a lot-specific certificate. You mark the certificate as something you will ask the quotation to confirm, because a mention on a marketing PDF is not yet the certificate.

The decision is now visible. Sheet B is the candidate to specify, with the residue, the lysate matrix, the load, the treated-versus-untreated control, the freeze-thaw limit, and a request that the quotation agree to a lot certificate. Sheet A is not rejected forever. It is rejected for this matrix until the seller fills the empty cells with evidence that matches lysates. You did not paste either blot into the sourcing file.

If a third sheet arrives later, you add a column. You do not rebuild the rows around the newcomer's favourite heading.

The branch when a cell is empty

An empty cell is a stop, not a zero. Zero would mean the manufacturer measured nothing and reported nothing, which is a result. Empty means you did not find the field.

Ask, in writing, for that field. If the answer fills it with a different matrix or a different control, you have learned the product does not match. If the answer is a new PDF that still lacks the field, the cell stays empty and the product stays off the mandatory list. Do not average Sheet A's picture with Sheet B's clone and call the blend a specification.

If units conflict and you cannot convert exactly, keep both units and send the question with the enquiry. Inventing a factor so the column looks tidy is how a later certificate surprises you.

Failure modes after the table exists

The first is false precision. You fill a blank with a number from a review article or from the other sheet. The table then looks complete and is no longer yours. Put "not stated" in the cell and send the question.

The second is comparing unlike controls and declaring a winner. A peptide block and a genetic knockout answer different doubts. If your doubt is cross-reactivity in lysate, the knockout or a second independent antibody matters more than a block of the peptide you already believe. Choose the field that matches the doubt. Do not let the longer PDF win.

The third is stopping at the datasheet. The table selects what the quotation must promise. On receipt you still match the lot certificate to the cells you cared about. A datasheet from last year is not evidence about the vial in the box. ISO 9001 is only a public nudge toward traceable requirements. Your certificate match is the actual check.

Safety and research use

Extract the safety data sheet as a document field when the product is a chemical or a kit with hazards. The institution decides storage, waste, and containment. A datasheet claim about research use is not a diagnostic approval, and a strong blot is not a clinical validation. If the product is an infectious material or a toxin, stop and use the institutional biosafety route before you compare sensitivity figures.

Using the table in a warm, shared office

Print the laboratory table, not the two PDFs, for the meeting. In a humid week, ink on a cheap printout can smear; keep the file itself as the record, with the date you read each sheet and the sheet's revision if it has one. Do not add a city, a courier story, or a guessed transit time to the storage row. Storage is the class printed on the sheet. Whether a seller can meet that class on arrival is a separate written line in the quotation.

What the enquiry then asks

Send your table's chosen column as numbered lines: analyte, matrix, range, the control you require evidence of, storage class, and the documents the quotation must agree to supply. Ask the seller to fill any cell you left blank, in writing, without sending you a second brochure in their own order. Route it through laboratory procurement and, for a university group, academic research. Use contact or a quotation request. A scope can be discussed. Score the answer against your rows. If they send a new datasheet, extract it again. Do not let their layout become the specification.